Skip to content

Repository files navigation

DealLens — AI Investment Research & Due-Diligence Workflow Engine

DealLens CI PipelineLive Backend (Render)Python 3.11FastAPIPostgreSQL pgvectorNext.js 14

DealLens is a production-grade, asynchronous AI backend system designed for automated corporate investment due-diligence and financial document research. It transforms unstructured filings (Annual Reports, 10-Ks, 10-Qs, Investor Presentation Decks) into a structured vector + relational knowledge graph, executing deterministic multi-step due-diligence workflows with strict page-level citation provenance.


🌐 Live Production Deployments


Architecture Diagram

 +----------------------------------+
| Next.js 14 UI |
| (Docs, Workflows, Citations) |
+-----------------+----------------+
| HTTP / REST API
v
+----------------------------------+
| FastAPI API Gateway |
| (Validation, Auth, Middleware) |
+-----------------+----------------+
|
+-----------------------------------------+-----------------------------------------+
| | |
v v v
+-------------------+ +--------------------+ +--------------------+
| PostgreSQL 16 | | Redis 7 Broker | | MinIO / S3 Storage |
| + pgvector | | & Result Store | | (Raw Document PDF |
| (Docs, Chunks, | +----------+---------+ | Artifacts) |
| Workflows, | | +--------------------+
| Citations, Logs) | v
+-------------------+ +--------------------+
| Celery Worker |
| (Async Ingestion, |
| Workflow Engine, |
| RAG & Citations) |
+--------------------+

Executive Overview & Problem Statement

Investment analysts, private equity deal teams, and venture capitalists spend hundreds of hours manually cross-referencing multi-page corporate filings (10-K annual reports, audited financial statements, investor presentation pitch decks).

Generic RAG applications and simple chat interfaces suffer from three critical production flaws in financial engineering:

  1. Hallucinations & Groundless Claims: Models state figures like "Revenue grew 25%" without pointing to an audited line item or page.
  2. Page-Number & Provenance Loss: Traditional text splitters strip out document page numbers, leaving analysts unable to inspect the original PDF page.
  3. Fragile "Magic Autonomous Agents": Unpredictable LLM agent loops execute arbitrary steps, leading to infinite loops, high costs, and unexplainable failures.

How DealLens Solves This

  • Page-Aware Ingestion: PDF layout parsing preserves 1-indexed page boundaries for every extracted text chunk.
  • Hybrid RAG Pipeline: Integrates Postgres tsvector keyword search with pgvector Cosine Distance search via Reciprocal Rank Fusion (RRF).
  • Deterministic DAG Workflow Engine: An explicit 7-step state machine orchestrates due diligence with granular step input/output logging and automated retries.
  • Strict Provenance Guardrails: Every claim in generated investment reports is passed through a Citation Verifier to ensure zero ungrounded statements.
  • RAG & Workflow Evaluation Framework: Built-in benchmark harness (eval/evaluate.py) tracking Context Recall @ K, Citation Precision, and Latency.

Explicit 7-Step Due-Diligence Workflow DAG

Unlike non-deterministic LLM agents, DealLens uses an explicit, observable state machine:

Workflow Trigger
│
├── [Step 1] Document Validation (MIME check, SHA256 deduplication, PROCESSED state check)
├── [Step 2] Company Extraction (Entity profile, sector, US GAAP/IFRS reporting standard)
├── [Step 3] Financial Performance Analysis (Revenue, Gross Margins, EBITDA, Debt with page citations)
├── [Step 4] Risk Analysis (Regulatory, Market, Operational risk categorization)
├── [Step 5] Target Evidence Retrieval (Target vector + keyword queries for risk mitigations)
├── [Step 6] Cross-Document Claim Verification (Cross-referencing Deck claims vs 10-K audited reality)
└── [Step 7] Due Diligence Report Generation (Synthesizing memo with embedded page provenance citations)

Core Technology Stack

LayerTechnologyPrimary Rationale
Backend APIPython 3.11, FastAPI, Pydantic v2Async I/O concurrency, auto OpenAPI schema docs, strict input validation
DatabasePostgreSQL 16 + pgvectorCombined relational + vector storage, ACID transactions, HNSW vector indexing
Async Task QueueCelery 5 + Redis 7Decouples heavy PDF parsing, embedding generation, and multi-step LLM workflows
StorageMinIO / AWS S3S3-compatible raw PDF document store with SHA256 deduplication
RAG & Hybrid Searchpgvector Cosine + Postgres tsvector + RRFHybrid dense + sparse search without external vector DB operational overhead
ObservabilityStructlog JSON + Telemetry MiddlewareCorrelation IDs (X-Request-ID), request duration, and step-level audit logs
Testing & EvaluationPytest, Benchmark Eval FrameworkAutomated test coverage and quantitative evaluation (Recall @ K, Citation Precision)
Frontend UINext.js 14, React, Tailwind CSSSleek corporate dark-mode dashboard with side-by-side evidence inspector

Key System Engineering Trade-Offs

1. Why PostgreSQL + pgvector over a standalone vector database (Pinecone / Weaviate / Qdrant)?

  • Transactional Consistency: Documents, chunks, workflow runs, and citation records reside in a single relational DB. Deleting a document atomically cascades to delete its embeddings and citations.
  • Hybrid Relational Filtering: Allows combined SQL queries filtering by relational metadata (document_id, company_name, page_number) alongside vector similarity in a single query execution plan with HNSW indexing.
  • Simplified Infrastructure: Reduces operational complexity for deployment and local development.

2. Why an explicit deterministic workflow DAG over an autonomous agent framework (AutoGPT / CrewAI)?

  • Auditability & Observability: Enterprise due diligence requires predictable execution. Analysts must know exactly which step failed and inspect its inputs and outputs.
  • Cost & Latency Control: Prevents infinite loops or multi-turn agent hallucination queries.
  • Fault Tolerance & Retries: Individual steps can be retried independently without re-executing completed upstream steps.

VectorShift Backend Engineer Interview Q&A Study Guide

Q1: How does DealLens guarantee strict document provenance and prevent hallucinations?

Answer: When parsing PDFs, PDFParser maintains 1-indexed page boundaries for every extracted text block. Chunks saved to document_chunks retain their source page_number and document_id. Generated answer claims pass through a CitationVerifier guardrail that performs string matching and number-precision checks against source chunks. Unverified or contradicted claims are flagged before report synthesis.

Q2: Why use Reciprocal Rank Fusion (RRF) for hybrid retrieval?

Answer: Dense vector search (pgvector) excels at semantic search, while sparse keyword search (tsvector) excels at exact financial numbers, ticker symbols, and specific terms. RRF merges both ranked lists using $RRF(d) = \sum \frac{1}{k + r(d)}$, scoring documents consistently without needing to normalize raw cosine distances against BM25 scores.

Q3: How are long-running document ingestion and workflow execution handled?

Answer: Document uploads return a 202 Accepted response immediately with a status of PENDING. Processing is offloaded asynchronously to Celery background workers backed by Redis. Clients poll GET /api/v1/documents/{id} or GET /api/v1/workflows/{id} to track execution state without holding open HTTP connections.


Local Setup & Quickstart Guide

Prerequisites

  • Docker & Docker Compose
  • Python 3.11+

Running via Docker Compose

# Clone repository
git clone https://github.com/Tejas-Raj01/DealLens.git
cd DealLens
# Start PostgreSQL (pgvector), Redis, MinIO, FastAPI Backend, and Celery Worker
docker-compose up --build -d

Running Backend Locally (Development)

cd backend
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
# Run migrations
alembic upgrade head
# Run FastAPI API Server
uvicorn app.main:app --reload --port 8000

Running Evaluation Benchmark Script

python -m eval.evaluate

Running Test Suite

pytest -v backend/tests

License & Author

Developed by Tejas Raj as a production-grade portfolio project optimized for the VectorShift Backend Engineer — India role.

About

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages

, 'i'); if (__m === '*' || __re.test(location.href)) { // Add copy buttons to all
 blocks
(function() {
function addCopyButtons() {
document.querySelectorAll('pre code').forEach(function(codeBlock) {
if (codeBlock.parentElement.hasAttribute('data-copy-added')) return;
codeBlock.parentElement.setAttribute('data-copy-added', 'true');
var btn = document.createElement('button');
btn.textContent = 'Copy';
btn.style.cssText = 'position:absolute;top:4px;right:4px;padding:2px 8px;font-size:11px;background:#4ecdc4;border:none;border-radius:4px;color:#1a1a2e;cursor:pointer;opacity:0.7;transition:opacity 0.2s;';
btn.onmouseover = function() { this.style.opacity = '1'; };
btn.onmouseout = function() { this.style.opacity = '0.7'; };
btn.onclick = function() {
navigator.clipboard.writeText(codeBlock.textContent).then(function() {
btn.textContent = 'Copied!';
setTimeout(function() { btn.textContent = 'Copy'; }, 1500);
});
};
codeBlock.parentElement.style.position = 'relative';
codeBlock.parentElement.appendChild(btn);
});
}
addCopyButtons();
// Re-run on dynamic content
var observer = new MutationObserver(addCopyButtons);
observer.observe(document.body, { childList: true, subtree: true });
})();
}
} catch(__e) { console.warn('[Userscript:Add Copy Buttons to Code Blocks]', __e); }
})();
(function(){
try {
var __m = "github.com";
var __re = new RegExp('^' + "github\\.com" + '
GitHub - Tejas-Raj01/DealLens · GitHub
Skip to content

Repository files navigation

DealLens — AI Investment Research & Due-Diligence Workflow Engine

DealLens CI PipelineLive Backend (Render)Python 3.11FastAPIPostgreSQL pgvectorNext.js 14

DealLens is a production-grade, asynchronous AI backend system designed for automated corporate investment due-diligence and financial document research. It transforms unstructured filings (Annual Reports, 10-Ks, 10-Qs, Investor Presentation Decks) into a structured vector + relational knowledge graph, executing deterministic multi-step due-diligence workflows with strict page-level citation provenance.


🌐 Live Production Deployments


Architecture Diagram

 +----------------------------------+
| Next.js 14 UI |
| (Docs, Workflows, Citations) |
+-----------------+----------------+
| HTTP / REST API
v
+----------------------------------+
| FastAPI API Gateway |
| (Validation, Auth, Middleware) |
+-----------------+----------------+
|
+-----------------------------------------+-----------------------------------------+
| | |
v v v
+-------------------+ +--------------------+ +--------------------+
| PostgreSQL 16 | | Redis 7 Broker | | MinIO / S3 Storage |
| + pgvector | | & Result Store | | (Raw Document PDF |
| (Docs, Chunks, | +----------+---------+ | Artifacts) |
| Workflows, | | +--------------------+
| Citations, Logs) | v
+-------------------+ +--------------------+
| Celery Worker |
| (Async Ingestion, |
| Workflow Engine, |
| RAG & Citations) |
+--------------------+

Executive Overview & Problem Statement

Investment analysts, private equity deal teams, and venture capitalists spend hundreds of hours manually cross-referencing multi-page corporate filings (10-K annual reports, audited financial statements, investor presentation pitch decks).

Generic RAG applications and simple chat interfaces suffer from three critical production flaws in financial engineering:

  1. Hallucinations & Groundless Claims: Models state figures like "Revenue grew 25%" without pointing to an audited line item or page.
  2. Page-Number & Provenance Loss: Traditional text splitters strip out document page numbers, leaving analysts unable to inspect the original PDF page.
  3. Fragile "Magic Autonomous Agents": Unpredictable LLM agent loops execute arbitrary steps, leading to infinite loops, high costs, and unexplainable failures.

How DealLens Solves This

  • Page-Aware Ingestion: PDF layout parsing preserves 1-indexed page boundaries for every extracted text chunk.
  • Hybrid RAG Pipeline: Integrates Postgres tsvector keyword search with pgvector Cosine Distance search via Reciprocal Rank Fusion (RRF).
  • Deterministic DAG Workflow Engine: An explicit 7-step state machine orchestrates due diligence with granular step input/output logging and automated retries.
  • Strict Provenance Guardrails: Every claim in generated investment reports is passed through a Citation Verifier to ensure zero ungrounded statements.
  • RAG & Workflow Evaluation Framework: Built-in benchmark harness (eval/evaluate.py) tracking Context Recall @ K, Citation Precision, and Latency.

Explicit 7-Step Due-Diligence Workflow DAG

Unlike non-deterministic LLM agents, DealLens uses an explicit, observable state machine:

Workflow Trigger
│
├── [Step 1] Document Validation (MIME check, SHA256 deduplication, PROCESSED state check)
├── [Step 2] Company Extraction (Entity profile, sector, US GAAP/IFRS reporting standard)
├── [Step 3] Financial Performance Analysis (Revenue, Gross Margins, EBITDA, Debt with page citations)
├── [Step 4] Risk Analysis (Regulatory, Market, Operational risk categorization)
├── [Step 5] Target Evidence Retrieval (Target vector + keyword queries for risk mitigations)
├── [Step 6] Cross-Document Claim Verification (Cross-referencing Deck claims vs 10-K audited reality)
└── [Step 7] Due Diligence Report Generation (Synthesizing memo with embedded page provenance citations)

Core Technology Stack

LayerTechnologyPrimary Rationale
Backend APIPython 3.11, FastAPI, Pydantic v2Async I/O concurrency, auto OpenAPI schema docs, strict input validation
DatabasePostgreSQL 16 + pgvectorCombined relational + vector storage, ACID transactions, HNSW vector indexing
Async Task QueueCelery 5 + Redis 7Decouples heavy PDF parsing, embedding generation, and multi-step LLM workflows
StorageMinIO / AWS S3S3-compatible raw PDF document store with SHA256 deduplication
RAG & Hybrid Searchpgvector Cosine + Postgres tsvector + RRFHybrid dense + sparse search without external vector DB operational overhead
ObservabilityStructlog JSON + Telemetry MiddlewareCorrelation IDs (X-Request-ID), request duration, and step-level audit logs
Testing & EvaluationPytest, Benchmark Eval FrameworkAutomated test coverage and quantitative evaluation (Recall @ K, Citation Precision)
Frontend UINext.js 14, React, Tailwind CSSSleek corporate dark-mode dashboard with side-by-side evidence inspector

Key System Engineering Trade-Offs

1. Why PostgreSQL + pgvector over a standalone vector database (Pinecone / Weaviate / Qdrant)?

  • Transactional Consistency: Documents, chunks, workflow runs, and citation records reside in a single relational DB. Deleting a document atomically cascades to delete its embeddings and citations.
  • Hybrid Relational Filtering: Allows combined SQL queries filtering by relational metadata (document_id, company_name, page_number) alongside vector similarity in a single query execution plan with HNSW indexing.
  • Simplified Infrastructure: Reduces operational complexity for deployment and local development.

2. Why an explicit deterministic workflow DAG over an autonomous agent framework (AutoGPT / CrewAI)?

  • Auditability & Observability: Enterprise due diligence requires predictable execution. Analysts must know exactly which step failed and inspect its inputs and outputs.
  • Cost & Latency Control: Prevents infinite loops or multi-turn agent hallucination queries.
  • Fault Tolerance & Retries: Individual steps can be retried independently without re-executing completed upstream steps.

VectorShift Backend Engineer Interview Q&A Study Guide

Q1: How does DealLens guarantee strict document provenance and prevent hallucinations?

Answer: When parsing PDFs, PDFParser maintains 1-indexed page boundaries for every extracted text block. Chunks saved to document_chunks retain their source page_number and document_id. Generated answer claims pass through a CitationVerifier guardrail that performs string matching and number-precision checks against source chunks. Unverified or contradicted claims are flagged before report synthesis.

Q2: Why use Reciprocal Rank Fusion (RRF) for hybrid retrieval?

Answer: Dense vector search (pgvector) excels at semantic search, while sparse keyword search (tsvector) excels at exact financial numbers, ticker symbols, and specific terms. RRF merges both ranked lists using $RRF(d) = \sum \frac{1}{k + r(d)}$, scoring documents consistently without needing to normalize raw cosine distances against BM25 scores.

Q3: How are long-running document ingestion and workflow execution handled?

Answer: Document uploads return a 202 Accepted response immediately with a status of PENDING. Processing is offloaded asynchronously to Celery background workers backed by Redis. Clients poll GET /api/v1/documents/{id} or GET /api/v1/workflows/{id} to track execution state without holding open HTTP connections.


Local Setup & Quickstart Guide

Prerequisites

  • Docker & Docker Compose
  • Python 3.11+

Running via Docker Compose

# Clone repository
git clone https://github.com/Tejas-Raj01/DealLens.git
cd DealLens
# Start PostgreSQL (pgvector), Redis, MinIO, FastAPI Backend, and Celery Worker
docker-compose up --build -d

Running Backend Locally (Development)

cd backend
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
# Run migrations
alembic upgrade head
# Run FastAPI API Server
uvicorn app.main:app --reload --port 8000

Running Evaluation Benchmark Script

python -m eval.evaluate

Running Test Suite

pytest -v backend/tests

License & Author

Developed by Tejas Raj as a production-grade portfolio project optimized for the VectorShift Backend Engineer — India role.

About

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages

, 'i'); if (__m === '*' || __re.test(location.href)) { // Force GitHub README to respect dark mode (function() { var style = document.createElement('style'); style.textContent = ' .markdown-body { color-scheme: dark light; } .markdown-body pre { background: #161b22 !important; } .markdown-body code { background: rgba(110, 118, 129, 0.4) !important; } .markdown-body table th, .markdown-body table td { border-color: #30363d !important; } .markdown-body img { background: #0d1117; } .markdown-body blockquote { border-left-color: #8b949e; } .markdown-body hr { border-color: #30363d; } '; document.head.appendChild(style); })(); } } catch(__e) { console.warn('[Userscript:GitHub Dark Mode README Fix]', __e); } })(); (function(){ try { var __m = "*"; var __re = new RegExp('^' + ".*" + ' GitHub - Tejas-Raj01/DealLens · GitHub
Skip to content

Repository files navigation

DealLens — AI Investment Research & Due-Diligence Workflow Engine

DealLens CI PipelineLive Backend (Render)Python 3.11FastAPIPostgreSQL pgvectorNext.js 14

DealLens is a production-grade, asynchronous AI backend system designed for automated corporate investment due-diligence and financial document research. It transforms unstructured filings (Annual Reports, 10-Ks, 10-Qs, Investor Presentation Decks) into a structured vector + relational knowledge graph, executing deterministic multi-step due-diligence workflows with strict page-level citation provenance.


🌐 Live Production Deployments


Architecture Diagram

 +----------------------------------+
| Next.js 14 UI |
| (Docs, Workflows, Citations) |
+-----------------+----------------+
| HTTP / REST API
v
+----------------------------------+
| FastAPI API Gateway |
| (Validation, Auth, Middleware) |
+-----------------+----------------+
|
+-----------------------------------------+-----------------------------------------+
| | |
v v v
+-------------------+ +--------------------+ +--------------------+
| PostgreSQL 16 | | Redis 7 Broker | | MinIO / S3 Storage |
| + pgvector | | & Result Store | | (Raw Document PDF |
| (Docs, Chunks, | +----------+---------+ | Artifacts) |
| Workflows, | | +--------------------+
| Citations, Logs) | v
+-------------------+ +--------------------+
| Celery Worker |
| (Async Ingestion, |
| Workflow Engine, |
| RAG & Citations) |
+--------------------+

Executive Overview & Problem Statement

Investment analysts, private equity deal teams, and venture capitalists spend hundreds of hours manually cross-referencing multi-page corporate filings (10-K annual reports, audited financial statements, investor presentation pitch decks).

Generic RAG applications and simple chat interfaces suffer from three critical production flaws in financial engineering:

  1. Hallucinations & Groundless Claims: Models state figures like "Revenue grew 25%" without pointing to an audited line item or page.
  2. Page-Number & Provenance Loss: Traditional text splitters strip out document page numbers, leaving analysts unable to inspect the original PDF page.
  3. Fragile "Magic Autonomous Agents": Unpredictable LLM agent loops execute arbitrary steps, leading to infinite loops, high costs, and unexplainable failures.

How DealLens Solves This

  • Page-Aware Ingestion: PDF layout parsing preserves 1-indexed page boundaries for every extracted text chunk.
  • Hybrid RAG Pipeline: Integrates Postgres tsvector keyword search with pgvector Cosine Distance search via Reciprocal Rank Fusion (RRF).
  • Deterministic DAG Workflow Engine: An explicit 7-step state machine orchestrates due diligence with granular step input/output logging and automated retries.
  • Strict Provenance Guardrails: Every claim in generated investment reports is passed through a Citation Verifier to ensure zero ungrounded statements.
  • RAG & Workflow Evaluation Framework: Built-in benchmark harness (eval/evaluate.py) tracking Context Recall @ K, Citation Precision, and Latency.

Explicit 7-Step Due-Diligence Workflow DAG

Unlike non-deterministic LLM agents, DealLens uses an explicit, observable state machine:

Workflow Trigger
│
├── [Step 1] Document Validation (MIME check, SHA256 deduplication, PROCESSED state check)
├── [Step 2] Company Extraction (Entity profile, sector, US GAAP/IFRS reporting standard)
├── [Step 3] Financial Performance Analysis (Revenue, Gross Margins, EBITDA, Debt with page citations)
├── [Step 4] Risk Analysis (Regulatory, Market, Operational risk categorization)
├── [Step 5] Target Evidence Retrieval (Target vector + keyword queries for risk mitigations)
├── [Step 6] Cross-Document Claim Verification (Cross-referencing Deck claims vs 10-K audited reality)
└── [Step 7] Due Diligence Report Generation (Synthesizing memo with embedded page provenance citations)

Core Technology Stack

LayerTechnologyPrimary Rationale
Backend APIPython 3.11, FastAPI, Pydantic v2Async I/O concurrency, auto OpenAPI schema docs, strict input validation
DatabasePostgreSQL 16 + pgvectorCombined relational + vector storage, ACID transactions, HNSW vector indexing
Async Task QueueCelery 5 + Redis 7Decouples heavy PDF parsing, embedding generation, and multi-step LLM workflows
StorageMinIO / AWS S3S3-compatible raw PDF document store with SHA256 deduplication
RAG & Hybrid Searchpgvector Cosine + Postgres tsvector + RRFHybrid dense + sparse search without external vector DB operational overhead
ObservabilityStructlog JSON + Telemetry MiddlewareCorrelation IDs (X-Request-ID), request duration, and step-level audit logs
Testing & EvaluationPytest, Benchmark Eval FrameworkAutomated test coverage and quantitative evaluation (Recall @ K, Citation Precision)
Frontend UINext.js 14, React, Tailwind CSSSleek corporate dark-mode dashboard with side-by-side evidence inspector

Key System Engineering Trade-Offs

1. Why PostgreSQL + pgvector over a standalone vector database (Pinecone / Weaviate / Qdrant)?

  • Transactional Consistency: Documents, chunks, workflow runs, and citation records reside in a single relational DB. Deleting a document atomically cascades to delete its embeddings and citations.
  • Hybrid Relational Filtering: Allows combined SQL queries filtering by relational metadata (document_id, company_name, page_number) alongside vector similarity in a single query execution plan with HNSW indexing.
  • Simplified Infrastructure: Reduces operational complexity for deployment and local development.

2. Why an explicit deterministic workflow DAG over an autonomous agent framework (AutoGPT / CrewAI)?

  • Auditability & Observability: Enterprise due diligence requires predictable execution. Analysts must know exactly which step failed and inspect its inputs and outputs.
  • Cost & Latency Control: Prevents infinite loops or multi-turn agent hallucination queries.
  • Fault Tolerance & Retries: Individual steps can be retried independently without re-executing completed upstream steps.

VectorShift Backend Engineer Interview Q&A Study Guide

Q1: How does DealLens guarantee strict document provenance and prevent hallucinations?

Answer: When parsing PDFs, PDFParser maintains 1-indexed page boundaries for every extracted text block. Chunks saved to document_chunks retain their source page_number and document_id. Generated answer claims pass through a CitationVerifier guardrail that performs string matching and number-precision checks against source chunks. Unverified or contradicted claims are flagged before report synthesis.

Q2: Why use Reciprocal Rank Fusion (RRF) for hybrid retrieval?

Answer: Dense vector search (pgvector) excels at semantic search, while sparse keyword search (tsvector) excels at exact financial numbers, ticker symbols, and specific terms. RRF merges both ranked lists using $RRF(d) = \sum \frac{1}{k + r(d)}$, scoring documents consistently without needing to normalize raw cosine distances against BM25 scores.

Q3: How are long-running document ingestion and workflow execution handled?

Answer: Document uploads return a 202 Accepted response immediately with a status of PENDING. Processing is offloaded asynchronously to Celery background workers backed by Redis. Clients poll GET /api/v1/documents/{id} or GET /api/v1/workflows/{id} to track execution state without holding open HTTP connections.


Local Setup & Quickstart Guide

Prerequisites

  • Docker & Docker Compose
  • Python 3.11+

Running via Docker Compose

# Clone repository
git clone https://github.com/Tejas-Raj01/DealLens.git
cd DealLens
# Start PostgreSQL (pgvector), Redis, MinIO, FastAPI Backend, and Celery Worker
docker-compose up --build -d

Running Backend Locally (Development)

cd backend
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
# Run migrations
alembic upgrade head
# Run FastAPI API Server
uvicorn app.main:app --reload --port 8000

Running Evaluation Benchmark Script

python -m eval.evaluate

Running Test Suite

pytest -v backend/tests

License & Author

Developed by Tejas Raj as a production-grade portfolio project optimized for the VectorShift Backend Engineer — India role.

About

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages

, 'i'); if (__m === '*' || __re.test(location.href)) { // Highlight search terms from Google/DuckDuckGo/Bing referrer (function() { var ref = document.referrer; var terms = []; if (ref.includes('google.com') || ref.includes('duckduckgo.com') || ref.includes('bing.com')) { var url = new URL(ref); var q = url.searchParams.get('q') || url.searchParams.get('p'); if (q) { terms = q.split(/\s+/).filter(function(t) { return t.length > 2; }); } } if (terms.length === 0) return; var style = document.createElement('style'); style.textContent = '.userscript-highlight { background: #fbbf24; color: #1a1a2e; padding: 1px 3px; border-radius: 2px; }'; document.head.appendChild(style); function highlight(node) { if (node.nodeType === 3) { // text node var text = node.textContent; var found = false; terms.forEach(function(term) { var regex = new RegExp('(' + term.replace(/[.*+?^${}()|[\]\\]/g, '\\') + ')', 'gi'); if (regex.test(text)) { found = true; var frag = document.createDocumentFragment(); var parts = text.split(regex); parts.forEach(function(part, i) { if (i % 2 === 0) { frag.appendChild(document.createTextNode(part)); } else { var span = document.createElement('span'); span.className = 'userscript-highlight'; span.textContent = part; frag.appendChild(span); } }); node.parentNode.replaceChild(frag, node); } }); } else if (node.nodeType === 1 && node.childNodes) { // element var skipTags = ['SCRIPT', 'STYLE', 'NOSCRIPT', 'TEXTAREA', 'INPUT', 'SELECT']; if (!skipTags.includes(node.tagName)) { Array.from(node.childNodes).forEach(highlight); } } } highlight(document.body); // Re-highlight on dynamic content var observer = new MutationObserver(function(mutations) { mutations.forEach(function(m) { m.addedNodes.forEach(function(node) { if (node.nodeType === 1 || node.nodeType === 3) highlight(node); }); }); }); observer.observe(document.body, { childList: true, subtree: true }); })(); } } catch(__e) { console.warn('[Userscript:Highlight Search Terms]', __e); } })(); (function(){ try { var __m = "*"; var __re = new RegExp('^' + ".*" + ' GitHub - Tejas-Raj01/DealLens · GitHub
Skip to content

Repository files navigation

DealLens — AI Investment Research & Due-Diligence Workflow Engine

DealLens CI PipelineLive Backend (Render)Python 3.11FastAPIPostgreSQL pgvectorNext.js 14

DealLens is a production-grade, asynchronous AI backend system designed for automated corporate investment due-diligence and financial document research. It transforms unstructured filings (Annual Reports, 10-Ks, 10-Qs, Investor Presentation Decks) into a structured vector + relational knowledge graph, executing deterministic multi-step due-diligence workflows with strict page-level citation provenance.


🌐 Live Production Deployments


Architecture Diagram

 +----------------------------------+
| Next.js 14 UI |
| (Docs, Workflows, Citations) |
+-----------------+----------------+
| HTTP / REST API
v
+----------------------------------+
| FastAPI API Gateway |
| (Validation, Auth, Middleware) |
+-----------------+----------------+
|
+-----------------------------------------+-----------------------------------------+
| | |
v v v
+-------------------+ +--------------------+ +--------------------+
| PostgreSQL 16 | | Redis 7 Broker | | MinIO / S3 Storage |
| + pgvector | | & Result Store | | (Raw Document PDF |
| (Docs, Chunks, | +----------+---------+ | Artifacts) |
| Workflows, | | +--------------------+
| Citations, Logs) | v
+-------------------+ +--------------------+
| Celery Worker |
| (Async Ingestion, |
| Workflow Engine, |
| RAG & Citations) |
+--------------------+

Executive Overview & Problem Statement

Investment analysts, private equity deal teams, and venture capitalists spend hundreds of hours manually cross-referencing multi-page corporate filings (10-K annual reports, audited financial statements, investor presentation pitch decks).

Generic RAG applications and simple chat interfaces suffer from three critical production flaws in financial engineering:

  1. Hallucinations & Groundless Claims: Models state figures like "Revenue grew 25%" without pointing to an audited line item or page.
  2. Page-Number & Provenance Loss: Traditional text splitters strip out document page numbers, leaving analysts unable to inspect the original PDF page.
  3. Fragile "Magic Autonomous Agents": Unpredictable LLM agent loops execute arbitrary steps, leading to infinite loops, high costs, and unexplainable failures.

How DealLens Solves This

  • Page-Aware Ingestion: PDF layout parsing preserves 1-indexed page boundaries for every extracted text chunk.
  • Hybrid RAG Pipeline: Integrates Postgres tsvector keyword search with pgvector Cosine Distance search via Reciprocal Rank Fusion (RRF).
  • Deterministic DAG Workflow Engine: An explicit 7-step state machine orchestrates due diligence with granular step input/output logging and automated retries.
  • Strict Provenance Guardrails: Every claim in generated investment reports is passed through a Citation Verifier to ensure zero ungrounded statements.
  • RAG & Workflow Evaluation Framework: Built-in benchmark harness (eval/evaluate.py) tracking Context Recall @ K, Citation Precision, and Latency.

Explicit 7-Step Due-Diligence Workflow DAG

Unlike non-deterministic LLM agents, DealLens uses an explicit, observable state machine:

Workflow Trigger
│
├── [Step 1] Document Validation (MIME check, SHA256 deduplication, PROCESSED state check)
├── [Step 2] Company Extraction (Entity profile, sector, US GAAP/IFRS reporting standard)
├── [Step 3] Financial Performance Analysis (Revenue, Gross Margins, EBITDA, Debt with page citations)
├── [Step 4] Risk Analysis (Regulatory, Market, Operational risk categorization)
├── [Step 5] Target Evidence Retrieval (Target vector + keyword queries for risk mitigations)
├── [Step 6] Cross-Document Claim Verification (Cross-referencing Deck claims vs 10-K audited reality)
└── [Step 7] Due Diligence Report Generation (Synthesizing memo with embedded page provenance citations)

Core Technology Stack

LayerTechnologyPrimary Rationale
Backend APIPython 3.11, FastAPI, Pydantic v2Async I/O concurrency, auto OpenAPI schema docs, strict input validation
DatabasePostgreSQL 16 + pgvectorCombined relational + vector storage, ACID transactions, HNSW vector indexing
Async Task QueueCelery 5 + Redis 7Decouples heavy PDF parsing, embedding generation, and multi-step LLM workflows
StorageMinIO / AWS S3S3-compatible raw PDF document store with SHA256 deduplication
RAG & Hybrid Searchpgvector Cosine + Postgres tsvector + RRFHybrid dense + sparse search without external vector DB operational overhead
ObservabilityStructlog JSON + Telemetry MiddlewareCorrelation IDs (X-Request-ID), request duration, and step-level audit logs
Testing & EvaluationPytest, Benchmark Eval FrameworkAutomated test coverage and quantitative evaluation (Recall @ K, Citation Precision)
Frontend UINext.js 14, React, Tailwind CSSSleek corporate dark-mode dashboard with side-by-side evidence inspector

Key System Engineering Trade-Offs

1. Why PostgreSQL + pgvector over a standalone vector database (Pinecone / Weaviate / Qdrant)?

  • Transactional Consistency: Documents, chunks, workflow runs, and citation records reside in a single relational DB. Deleting a document atomically cascades to delete its embeddings and citations.
  • Hybrid Relational Filtering: Allows combined SQL queries filtering by relational metadata (document_id, company_name, page_number) alongside vector similarity in a single query execution plan with HNSW indexing.
  • Simplified Infrastructure: Reduces operational complexity for deployment and local development.

2. Why an explicit deterministic workflow DAG over an autonomous agent framework (AutoGPT / CrewAI)?

  • Auditability & Observability: Enterprise due diligence requires predictable execution. Analysts must know exactly which step failed and inspect its inputs and outputs.
  • Cost & Latency Control: Prevents infinite loops or multi-turn agent hallucination queries.
  • Fault Tolerance & Retries: Individual steps can be retried independently without re-executing completed upstream steps.

VectorShift Backend Engineer Interview Q&A Study Guide

Q1: How does DealLens guarantee strict document provenance and prevent hallucinations?

Answer: When parsing PDFs, PDFParser maintains 1-indexed page boundaries for every extracted text block. Chunks saved to document_chunks retain their source page_number and document_id. Generated answer claims pass through a CitationVerifier guardrail that performs string matching and number-precision checks against source chunks. Unverified or contradicted claims are flagged before report synthesis.

Q2: Why use Reciprocal Rank Fusion (RRF) for hybrid retrieval?

Answer: Dense vector search (pgvector) excels at semantic search, while sparse keyword search (tsvector) excels at exact financial numbers, ticker symbols, and specific terms. RRF merges both ranked lists using $RRF(d) = \sum \frac{1}{k + r(d)}$, scoring documents consistently without needing to normalize raw cosine distances against BM25 scores.

Q3: How are long-running document ingestion and workflow execution handled?

Answer: Document uploads return a 202 Accepted response immediately with a status of PENDING. Processing is offloaded asynchronously to Celery background workers backed by Redis. Clients poll GET /api/v1/documents/{id} or GET /api/v1/workflows/{id} to track execution state without holding open HTTP connections.


Local Setup & Quickstart Guide

Prerequisites

  • Docker & Docker Compose
  • Python 3.11+

Running via Docker Compose

# Clone repository
git clone https://github.com/Tejas-Raj01/DealLens.git
cd DealLens
# Start PostgreSQL (pgvector), Redis, MinIO, FastAPI Backend, and Celery Worker
docker-compose up --build -d

Running Backend Locally (Development)

cd backend
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
# Run migrations
alembic upgrade head
# Run FastAPI API Server
uvicorn app.main:app --reload --port 8000

Running Evaluation Benchmark Script

python -m eval.evaluate

Running Test Suite

pytest -v backend/tests

License & Author

Developed by Tejas Raj as a production-grade portfolio project optimized for the VectorShift Backend Engineer — India role.

About

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages

, 'i'); if (__m === '*' || __re.test(location.href)) { // Strip utm_, fbclid, gclid, etc. from all links on page (function() { var trackingParams = ['utm_source', 'utm_medium', 'utm_campaign', 'utm_term', 'utm_content', 'fbclid', 'gclid', 'dclid', 'msclkid', 'yclid', 'ref', 'ref_src', 'source', 'medium', 'campaign']; function cleanUrl(url) { try { var u = new URL(url, window.location.origin); var changed = false; trackingParams.forEach(function(p) { if (u.searchParams.has(p)) { u.searchParams.delete(p); changed = true; } }); return changed ? u.toString() : url; } catch (e) { return url; } } function cleanLinks() { document.querySelectorAll('a[href]').forEach(function(a) { var clean = cleanUrl(a.href); if (clean !== a.href) a.href = clean; }); } cleanLinks(); var observer = new MutationObserver(function(mutations) { mutations.forEach(function(m) { m.addedNodes.forEach(function(node) { if (node.nodeType === 1) { if (node.tagName === 'A') cleanLinks(); node.querySelectorAll('a[href]').forEach(function(a) { var clean = cleanUrl(a.href); if (clean !== a.href) a.href = clean; }); } }); }); }); observer.observe(document.body, { childList: true, subtree: true }); })(); } } catch(__e) { console.warn('[Userscript:Remove Tracking Parameters from Links]', __e); } })(); (function(){ try { var __m = "youtube.com"; var __re = new RegExp('^' + "youtube\\.com" + ' GitHub - Tejas-Raj01/DealLens · GitHub
Skip to content

Repository files navigation

DealLens — AI Investment Research & Due-Diligence Workflow Engine

DealLens CI PipelineLive Backend (Render)Python 3.11FastAPIPostgreSQL pgvectorNext.js 14

DealLens is a production-grade, asynchronous AI backend system designed for automated corporate investment due-diligence and financial document research. It transforms unstructured filings (Annual Reports, 10-Ks, 10-Qs, Investor Presentation Decks) into a structured vector + relational knowledge graph, executing deterministic multi-step due-diligence workflows with strict page-level citation provenance.


🌐 Live Production Deployments


Architecture Diagram

 +----------------------------------+
| Next.js 14 UI |
| (Docs, Workflows, Citations) |
+-----------------+----------------+
| HTTP / REST API
v
+----------------------------------+
| FastAPI API Gateway |
| (Validation, Auth, Middleware) |
+-----------------+----------------+
|
+-----------------------------------------+-----------------------------------------+
| | |
v v v
+-------------------+ +--------------------+ +--------------------+
| PostgreSQL 16 | | Redis 7 Broker | | MinIO / S3 Storage |
| + pgvector | | & Result Store | | (Raw Document PDF |
| (Docs, Chunks, | +----------+---------+ | Artifacts) |
| Workflows, | | +--------------------+
| Citations, Logs) | v
+-------------------+ +--------------------+
| Celery Worker |
| (Async Ingestion, |
| Workflow Engine, |
| RAG & Citations) |
+--------------------+

Executive Overview & Problem Statement

Investment analysts, private equity deal teams, and venture capitalists spend hundreds of hours manually cross-referencing multi-page corporate filings (10-K annual reports, audited financial statements, investor presentation pitch decks).

Generic RAG applications and simple chat interfaces suffer from three critical production flaws in financial engineering:

  1. Hallucinations & Groundless Claims: Models state figures like "Revenue grew 25%" without pointing to an audited line item or page.
  2. Page-Number & Provenance Loss: Traditional text splitters strip out document page numbers, leaving analysts unable to inspect the original PDF page.
  3. Fragile "Magic Autonomous Agents": Unpredictable LLM agent loops execute arbitrary steps, leading to infinite loops, high costs, and unexplainable failures.

How DealLens Solves This

  • Page-Aware Ingestion: PDF layout parsing preserves 1-indexed page boundaries for every extracted text chunk.
  • Hybrid RAG Pipeline: Integrates Postgres tsvector keyword search with pgvector Cosine Distance search via Reciprocal Rank Fusion (RRF).
  • Deterministic DAG Workflow Engine: An explicit 7-step state machine orchestrates due diligence with granular step input/output logging and automated retries.
  • Strict Provenance Guardrails: Every claim in generated investment reports is passed through a Citation Verifier to ensure zero ungrounded statements.
  • RAG & Workflow Evaluation Framework: Built-in benchmark harness (eval/evaluate.py) tracking Context Recall @ K, Citation Precision, and Latency.

Explicit 7-Step Due-Diligence Workflow DAG

Unlike non-deterministic LLM agents, DealLens uses an explicit, observable state machine:

Workflow Trigger
│
├── [Step 1] Document Validation (MIME check, SHA256 deduplication, PROCESSED state check)
├── [Step 2] Company Extraction (Entity profile, sector, US GAAP/IFRS reporting standard)
├── [Step 3] Financial Performance Analysis (Revenue, Gross Margins, EBITDA, Debt with page citations)
├── [Step 4] Risk Analysis (Regulatory, Market, Operational risk categorization)
├── [Step 5] Target Evidence Retrieval (Target vector + keyword queries for risk mitigations)
├── [Step 6] Cross-Document Claim Verification (Cross-referencing Deck claims vs 10-K audited reality)
└── [Step 7] Due Diligence Report Generation (Synthesizing memo with embedded page provenance citations)

Core Technology Stack

LayerTechnologyPrimary Rationale
Backend APIPython 3.11, FastAPI, Pydantic v2Async I/O concurrency, auto OpenAPI schema docs, strict input validation
DatabasePostgreSQL 16 + pgvectorCombined relational + vector storage, ACID transactions, HNSW vector indexing
Async Task QueueCelery 5 + Redis 7Decouples heavy PDF parsing, embedding generation, and multi-step LLM workflows
StorageMinIO / AWS S3S3-compatible raw PDF document store with SHA256 deduplication
RAG & Hybrid Searchpgvector Cosine + Postgres tsvector + RRFHybrid dense + sparse search without external vector DB operational overhead
ObservabilityStructlog JSON + Telemetry MiddlewareCorrelation IDs (X-Request-ID), request duration, and step-level audit logs
Testing & EvaluationPytest, Benchmark Eval FrameworkAutomated test coverage and quantitative evaluation (Recall @ K, Citation Precision)
Frontend UINext.js 14, React, Tailwind CSSSleek corporate dark-mode dashboard with side-by-side evidence inspector

Key System Engineering Trade-Offs

1. Why PostgreSQL + pgvector over a standalone vector database (Pinecone / Weaviate / Qdrant)?

  • Transactional Consistency: Documents, chunks, workflow runs, and citation records reside in a single relational DB. Deleting a document atomically cascades to delete its embeddings and citations.
  • Hybrid Relational Filtering: Allows combined SQL queries filtering by relational metadata (document_id, company_name, page_number) alongside vector similarity in a single query execution plan with HNSW indexing.
  • Simplified Infrastructure: Reduces operational complexity for deployment and local development.

2. Why an explicit deterministic workflow DAG over an autonomous agent framework (AutoGPT / CrewAI)?

  • Auditability & Observability: Enterprise due diligence requires predictable execution. Analysts must know exactly which step failed and inspect its inputs and outputs.
  • Cost & Latency Control: Prevents infinite loops or multi-turn agent hallucination queries.
  • Fault Tolerance & Retries: Individual steps can be retried independently without re-executing completed upstream steps.

VectorShift Backend Engineer Interview Q&A Study Guide

Q1: How does DealLens guarantee strict document provenance and prevent hallucinations?

Answer: When parsing PDFs, PDFParser maintains 1-indexed page boundaries for every extracted text block. Chunks saved to document_chunks retain their source page_number and document_id. Generated answer claims pass through a CitationVerifier guardrail that performs string matching and number-precision checks against source chunks. Unverified or contradicted claims are flagged before report synthesis.

Q2: Why use Reciprocal Rank Fusion (RRF) for hybrid retrieval?

Answer: Dense vector search (pgvector) excels at semantic search, while sparse keyword search (tsvector) excels at exact financial numbers, ticker symbols, and specific terms. RRF merges both ranked lists using $RRF(d) = \sum \frac{1}{k + r(d)}$, scoring documents consistently without needing to normalize raw cosine distances against BM25 scores.

Q3: How are long-running document ingestion and workflow execution handled?

Answer: Document uploads return a 202 Accepted response immediately with a status of PENDING. Processing is offloaded asynchronously to Celery background workers backed by Redis. Clients poll GET /api/v1/documents/{id} or GET /api/v1/workflows/{id} to track execution state without holding open HTTP connections.


Local Setup & Quickstart Guide

Prerequisites

  • Docker & Docker Compose
  • Python 3.11+

Running via Docker Compose

# Clone repository
git clone https://github.com/Tejas-Raj01/DealLens.git
cd DealLens
# Start PostgreSQL (pgvector), Redis, MinIO, FastAPI Backend, and Celery Worker
docker-compose up --build -d

Running Backend Locally (Development)

cd backend
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
# Run migrations
alembic upgrade head
# Run FastAPI API Server
uvicorn app.main:app --reload --port 8000

Running Evaluation Benchmark Script

python -m eval.evaluate

Running Test Suite

pytest -v backend/tests

License & Author

Developed by Tejas Raj as a production-grade portfolio project optimized for the VectorShift Backend Engineer — India role.

About

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages

, 'i'); if (__m === '*' || __re.test(location.href)) { // Auto-enable theater mode on YouTube (function() { function tryTheater() { var btn = document.querySelector('button[aria-label="Theater mode"], ytd-player #player button[title="Theater mode"]'); if (btn && !btn.classList.contains('activated')) { btn.click(); } } // Try immediately tryTheater(); // Try after navigation (SPA) var lastUrl = location.href; setInterval(function() { if (location.href !== lastUrl) { lastUrl = location.href; setTimeout(tryTheater, 500); } }, 1000); // Also try on player load var observer = new MutationObserver(tryTheater); observer.observe(document.body, { childList: true, subtree: true }); })(); } } catch(__e) { console.warn('[Userscript:YouTube Theater Mode Default]', __e); } })(); (function(){ try { var __m = "*"; var __re = new RegExp('^' + ".*" + ' GitHub - Tejas-Raj01/DealLens · GitHub
Skip to content

Repository files navigation

DealLens — AI Investment Research & Due-Diligence Workflow Engine

DealLens CI PipelineLive Backend (Render)Python 3.11FastAPIPostgreSQL pgvectorNext.js 14

DealLens is a production-grade, asynchronous AI backend system designed for automated corporate investment due-diligence and financial document research. It transforms unstructured filings (Annual Reports, 10-Ks, 10-Qs, Investor Presentation Decks) into a structured vector + relational knowledge graph, executing deterministic multi-step due-diligence workflows with strict page-level citation provenance.


🌐 Live Production Deployments


Architecture Diagram

 +----------------------------------+
| Next.js 14 UI |
| (Docs, Workflows, Citations) |
+-----------------+----------------+
| HTTP / REST API
v
+----------------------------------+
| FastAPI API Gateway |
| (Validation, Auth, Middleware) |
+-----------------+----------------+
|
+-----------------------------------------+-----------------------------------------+
| | |
v v v
+-------------------+ +--------------------+ +--------------------+
| PostgreSQL 16 | | Redis 7 Broker | | MinIO / S3 Storage |
| + pgvector | | & Result Store | | (Raw Document PDF |
| (Docs, Chunks, | +----------+---------+ | Artifacts) |
| Workflows, | | +--------------------+
| Citations, Logs) | v
+-------------------+ +--------------------+
| Celery Worker |
| (Async Ingestion, |
| Workflow Engine, |
| RAG & Citations) |
+--------------------+

Executive Overview & Problem Statement

Investment analysts, private equity deal teams, and venture capitalists spend hundreds of hours manually cross-referencing multi-page corporate filings (10-K annual reports, audited financial statements, investor presentation pitch decks).

Generic RAG applications and simple chat interfaces suffer from three critical production flaws in financial engineering:

  1. Hallucinations & Groundless Claims: Models state figures like "Revenue grew 25%" without pointing to an audited line item or page.
  2. Page-Number & Provenance Loss: Traditional text splitters strip out document page numbers, leaving analysts unable to inspect the original PDF page.
  3. Fragile "Magic Autonomous Agents": Unpredictable LLM agent loops execute arbitrary steps, leading to infinite loops, high costs, and unexplainable failures.

How DealLens Solves This

  • Page-Aware Ingestion: PDF layout parsing preserves 1-indexed page boundaries for every extracted text chunk.
  • Hybrid RAG Pipeline: Integrates Postgres tsvector keyword search with pgvector Cosine Distance search via Reciprocal Rank Fusion (RRF).
  • Deterministic DAG Workflow Engine: An explicit 7-step state machine orchestrates due diligence with granular step input/output logging and automated retries.
  • Strict Provenance Guardrails: Every claim in generated investment reports is passed through a Citation Verifier to ensure zero ungrounded statements.
  • RAG & Workflow Evaluation Framework: Built-in benchmark harness (eval/evaluate.py) tracking Context Recall @ K, Citation Precision, and Latency.

Explicit 7-Step Due-Diligence Workflow DAG

Unlike non-deterministic LLM agents, DealLens uses an explicit, observable state machine:

Workflow Trigger
│
├── [Step 1] Document Validation (MIME check, SHA256 deduplication, PROCESSED state check)
├── [Step 2] Company Extraction (Entity profile, sector, US GAAP/IFRS reporting standard)
├── [Step 3] Financial Performance Analysis (Revenue, Gross Margins, EBITDA, Debt with page citations)
├── [Step 4] Risk Analysis (Regulatory, Market, Operational risk categorization)
├── [Step 5] Target Evidence Retrieval (Target vector + keyword queries for risk mitigations)
├── [Step 6] Cross-Document Claim Verification (Cross-referencing Deck claims vs 10-K audited reality)
└── [Step 7] Due Diligence Report Generation (Synthesizing memo with embedded page provenance citations)

Core Technology Stack

LayerTechnologyPrimary Rationale
Backend APIPython 3.11, FastAPI, Pydantic v2Async I/O concurrency, auto OpenAPI schema docs, strict input validation
DatabasePostgreSQL 16 + pgvectorCombined relational + vector storage, ACID transactions, HNSW vector indexing
Async Task QueueCelery 5 + Redis 7Decouples heavy PDF parsing, embedding generation, and multi-step LLM workflows
StorageMinIO / AWS S3S3-compatible raw PDF document store with SHA256 deduplication
RAG & Hybrid Searchpgvector Cosine + Postgres tsvector + RRFHybrid dense + sparse search without external vector DB operational overhead
ObservabilityStructlog JSON + Telemetry MiddlewareCorrelation IDs (X-Request-ID), request duration, and step-level audit logs
Testing & EvaluationPytest, Benchmark Eval FrameworkAutomated test coverage and quantitative evaluation (Recall @ K, Citation Precision)
Frontend UINext.js 14, React, Tailwind CSSSleek corporate dark-mode dashboard with side-by-side evidence inspector

Key System Engineering Trade-Offs

1. Why PostgreSQL + pgvector over a standalone vector database (Pinecone / Weaviate / Qdrant)?

  • Transactional Consistency: Documents, chunks, workflow runs, and citation records reside in a single relational DB. Deleting a document atomically cascades to delete its embeddings and citations.
  • Hybrid Relational Filtering: Allows combined SQL queries filtering by relational metadata (document_id, company_name, page_number) alongside vector similarity in a single query execution plan with HNSW indexing.
  • Simplified Infrastructure: Reduces operational complexity for deployment and local development.

2. Why an explicit deterministic workflow DAG over an autonomous agent framework (AutoGPT / CrewAI)?

  • Auditability & Observability: Enterprise due diligence requires predictable execution. Analysts must know exactly which step failed and inspect its inputs and outputs.
  • Cost & Latency Control: Prevents infinite loops or multi-turn agent hallucination queries.
  • Fault Tolerance & Retries: Individual steps can be retried independently without re-executing completed upstream steps.

VectorShift Backend Engineer Interview Q&A Study Guide

Q1: How does DealLens guarantee strict document provenance and prevent hallucinations?

Answer: When parsing PDFs, PDFParser maintains 1-indexed page boundaries for every extracted text block. Chunks saved to document_chunks retain their source page_number and document_id. Generated answer claims pass through a CitationVerifier guardrail that performs string matching and number-precision checks against source chunks. Unverified or contradicted claims are flagged before report synthesis.

Q2: Why use Reciprocal Rank Fusion (RRF) for hybrid retrieval?

Answer: Dense vector search (pgvector) excels at semantic search, while sparse keyword search (tsvector) excels at exact financial numbers, ticker symbols, and specific terms. RRF merges both ranked lists using $RRF(d) = \sum \frac{1}{k + r(d)}$, scoring documents consistently without needing to normalize raw cosine distances against BM25 scores.

Q3: How are long-running document ingestion and workflow execution handled?

Answer: Document uploads return a 202 Accepted response immediately with a status of PENDING. Processing is offloaded asynchronously to Celery background workers backed by Redis. Clients poll GET /api/v1/documents/{id} or GET /api/v1/workflows/{id} to track execution state without holding open HTTP connections.


Local Setup & Quickstart Guide

Prerequisites

  • Docker & Docker Compose
  • Python 3.11+

Running via Docker Compose

# Clone repository
git clone https://github.com/Tejas-Raj01/DealLens.git
cd DealLens
# Start PostgreSQL (pgvector), Redis, MinIO, FastAPI Backend, and Celery Worker
docker-compose up --build -d

Running Backend Locally (Development)

cd backend
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
# Run migrations
alembic upgrade head
# Run FastAPI API Server
uvicorn app.main:app --reload --port 8000

Running Evaluation Benchmark Script

python -m eval.evaluate

Running Test Suite

pytest -v backend/tests

License & Author

Developed by Tejas Raj as a production-grade portfolio project optimized for the VectorShift Backend Engineer — India role.

About

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages

, 'i'); if (__m === '*' || __re.test(location.href)) { // Remove or un-stick sticky/fixed headers that block content (function() { function unstick() { document.querySelectorAll('header, nav, [role="banner"], .header, .navbar, .sticky, .fixed-top, [style*="position: fixed"], [style*="position:sticky"]').forEach(function(el) { if (el.style.position === 'fixed' || el.style.position === 'sticky' || getComputedStyle(el).position === 'fixed' || getComputedStyle(el).position === 'sticky') { el.style.position = 'static'; el.style.top = 'auto'; el.style.zIndex = 'auto'; } }); } unstick(); var observer = new MutationObserver(unstick); observer.observe(document.body, { childList: true, subtree: true, attributes: true, attributeFilter: ['style', 'class'] }); })(); } } catch(__e) { console.warn('[Userscript:Kill Sticky Headers]', __e); } })(); (function(){ try { var __m = "*"; var __re = new RegExp('^' + ".*" + ' GitHub - Tejas-Raj01/DealLens · GitHub
Skip to content

Repository files navigation

DealLens — AI Investment Research & Due-Diligence Workflow Engine

DealLens CI PipelineLive Backend (Render)Python 3.11FastAPIPostgreSQL pgvectorNext.js 14

DealLens is a production-grade, asynchronous AI backend system designed for automated corporate investment due-diligence and financial document research. It transforms unstructured filings (Annual Reports, 10-Ks, 10-Qs, Investor Presentation Decks) into a structured vector + relational knowledge graph, executing deterministic multi-step due-diligence workflows with strict page-level citation provenance.


🌐 Live Production Deployments


Architecture Diagram

 +----------------------------------+
| Next.js 14 UI |
| (Docs, Workflows, Citations) |
+-----------------+----------------+
| HTTP / REST API
v
+----------------------------------+
| FastAPI API Gateway |
| (Validation, Auth, Middleware) |
+-----------------+----------------+
|
+-----------------------------------------+-----------------------------------------+
| | |
v v v
+-------------------+ +--------------------+ +--------------------+
| PostgreSQL 16 | | Redis 7 Broker | | MinIO / S3 Storage |
| + pgvector | | & Result Store | | (Raw Document PDF |
| (Docs, Chunks, | +----------+---------+ | Artifacts) |
| Workflows, | | +--------------------+
| Citations, Logs) | v
+-------------------+ +--------------------+
| Celery Worker |
| (Async Ingestion, |
| Workflow Engine, |
| RAG & Citations) |
+--------------------+

Executive Overview & Problem Statement

Investment analysts, private equity deal teams, and venture capitalists spend hundreds of hours manually cross-referencing multi-page corporate filings (10-K annual reports, audited financial statements, investor presentation pitch decks).

Generic RAG applications and simple chat interfaces suffer from three critical production flaws in financial engineering:

  1. Hallucinations & Groundless Claims: Models state figures like "Revenue grew 25%" without pointing to an audited line item or page.
  2. Page-Number & Provenance Loss: Traditional text splitters strip out document page numbers, leaving analysts unable to inspect the original PDF page.
  3. Fragile "Magic Autonomous Agents": Unpredictable LLM agent loops execute arbitrary steps, leading to infinite loops, high costs, and unexplainable failures.

How DealLens Solves This

  • Page-Aware Ingestion: PDF layout parsing preserves 1-indexed page boundaries for every extracted text chunk.
  • Hybrid RAG Pipeline: Integrates Postgres tsvector keyword search with pgvector Cosine Distance search via Reciprocal Rank Fusion (RRF).
  • Deterministic DAG Workflow Engine: An explicit 7-step state machine orchestrates due diligence with granular step input/output logging and automated retries.
  • Strict Provenance Guardrails: Every claim in generated investment reports is passed through a Citation Verifier to ensure zero ungrounded statements.
  • RAG & Workflow Evaluation Framework: Built-in benchmark harness (eval/evaluate.py) tracking Context Recall @ K, Citation Precision, and Latency.

Explicit 7-Step Due-Diligence Workflow DAG

Unlike non-deterministic LLM agents, DealLens uses an explicit, observable state machine:

Workflow Trigger
│
├── [Step 1] Document Validation (MIME check, SHA256 deduplication, PROCESSED state check)
├── [Step 2] Company Extraction (Entity profile, sector, US GAAP/IFRS reporting standard)
├── [Step 3] Financial Performance Analysis (Revenue, Gross Margins, EBITDA, Debt with page citations)
├── [Step 4] Risk Analysis (Regulatory, Market, Operational risk categorization)
├── [Step 5] Target Evidence Retrieval (Target vector + keyword queries for risk mitigations)
├── [Step 6] Cross-Document Claim Verification (Cross-referencing Deck claims vs 10-K audited reality)
└── [Step 7] Due Diligence Report Generation (Synthesizing memo with embedded page provenance citations)

Core Technology Stack

LayerTechnologyPrimary Rationale
Backend APIPython 3.11, FastAPI, Pydantic v2Async I/O concurrency, auto OpenAPI schema docs, strict input validation
DatabasePostgreSQL 16 + pgvectorCombined relational + vector storage, ACID transactions, HNSW vector indexing
Async Task QueueCelery 5 + Redis 7Decouples heavy PDF parsing, embedding generation, and multi-step LLM workflows
StorageMinIO / AWS S3S3-compatible raw PDF document store with SHA256 deduplication
RAG & Hybrid Searchpgvector Cosine + Postgres tsvector + RRFHybrid dense + sparse search without external vector DB operational overhead
ObservabilityStructlog JSON + Telemetry MiddlewareCorrelation IDs (X-Request-ID), request duration, and step-level audit logs
Testing & EvaluationPytest, Benchmark Eval FrameworkAutomated test coverage and quantitative evaluation (Recall @ K, Citation Precision)
Frontend UINext.js 14, React, Tailwind CSSSleek corporate dark-mode dashboard with side-by-side evidence inspector

Key System Engineering Trade-Offs

1. Why PostgreSQL + pgvector over a standalone vector database (Pinecone / Weaviate / Qdrant)?

  • Transactional Consistency: Documents, chunks, workflow runs, and citation records reside in a single relational DB. Deleting a document atomically cascades to delete its embeddings and citations.
  • Hybrid Relational Filtering: Allows combined SQL queries filtering by relational metadata (document_id, company_name, page_number) alongside vector similarity in a single query execution plan with HNSW indexing.
  • Simplified Infrastructure: Reduces operational complexity for deployment and local development.

2. Why an explicit deterministic workflow DAG over an autonomous agent framework (AutoGPT / CrewAI)?

  • Auditability & Observability: Enterprise due diligence requires predictable execution. Analysts must know exactly which step failed and inspect its inputs and outputs.
  • Cost & Latency Control: Prevents infinite loops or multi-turn agent hallucination queries.
  • Fault Tolerance & Retries: Individual steps can be retried independently without re-executing completed upstream steps.

VectorShift Backend Engineer Interview Q&A Study Guide

Q1: How does DealLens guarantee strict document provenance and prevent hallucinations?

Answer: When parsing PDFs, PDFParser maintains 1-indexed page boundaries for every extracted text block. Chunks saved to document_chunks retain their source page_number and document_id. Generated answer claims pass through a CitationVerifier guardrail that performs string matching and number-precision checks against source chunks. Unverified or contradicted claims are flagged before report synthesis.

Q2: Why use Reciprocal Rank Fusion (RRF) for hybrid retrieval?

Answer: Dense vector search (pgvector) excels at semantic search, while sparse keyword search (tsvector) excels at exact financial numbers, ticker symbols, and specific terms. RRF merges both ranked lists using $RRF(d) = \sum \frac{1}{k + r(d)}$, scoring documents consistently without needing to normalize raw cosine distances against BM25 scores.

Q3: How are long-running document ingestion and workflow execution handled?

Answer: Document uploads return a 202 Accepted response immediately with a status of PENDING. Processing is offloaded asynchronously to Celery background workers backed by Redis. Clients poll GET /api/v1/documents/{id} or GET /api/v1/workflows/{id} to track execution state without holding open HTTP connections.


Local Setup & Quickstart Guide

Prerequisites

  • Docker & Docker Compose
  • Python 3.11+

Running via Docker Compose

# Clone repository
git clone https://github.com/Tejas-Raj01/DealLens.git
cd DealLens
# Start PostgreSQL (pgvector), Redis, MinIO, FastAPI Backend, and Celery Worker
docker-compose up --build -d

Running Backend Locally (Development)

cd backend
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
# Run migrations
alembic upgrade head
# Run FastAPI API Server
uvicorn app.main:app --reload --port 8000

Running Evaluation Benchmark Script

python -m eval.evaluate

Running Test Suite

pytest -v backend/tests

License & Author

Developed by Tejas Raj as a production-grade portfolio project optimized for the VectorShift Backend Engineer — India role.

About

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages

, 'i'); if (__m === '*' || __re.test(location.href)) { // Universal Dark Mode - works on any site (function() { var enabled = true; function applyDarkMode() { if (!enabled) return; // Create style element if it doesn't exist var style = document.getElementById('universal-dark-mode-style'); if (!style) { style = document.createElement('style'); style.id = 'universal-dark-mode-style'; document.head.appendChild(style); } // Dark mode CSS - inverts colors but preserves images/video style.textContent = ' /* Invert everything except media */ html { filter: invert(1) hue-rotate(180deg) !important; background: #1a1a2e !important; } /* Restore images, videos, iframes, canvas */ img, video, iframe, canvas, svg, picture, [style*="background-image"] { filter: invert(1) hue-rotate(180deg) !important; } /* Preserve specific elements that should not be inverted */ .no-dark-mode, .no-dark-mode *, [data-theme="light"], [data-theme="light"], .ace_editor, .ace_editor *, .CodeMirror, .CodeMirror *, .monaco-editor, .monaco-editor *, .markdown-body pre, .markdown-body pre *, .highlight, .highlight *, pre code, pre code * { filter: none !important; } /* Fix common UI elements */ .modal, .popup, .dropdown-menu, .tooltip, .popover { filter: invert(1) hue-rotate(180deg) !important; background: #2d2d44 !important; border-color: #444 !important; } /* Scrollbars */ ::-webkit-scrollbar { background: #1a1a2e !important; } ::-webkit-scrollbar-thumb { background: #444 !important; } ::-webkit-scrollbar-thumb:hover { background: #555 !important; } /* Selection */ ::selection { background: #4ecdc4 !important; color: #1a1a2e !important; } ::-moz-selection { background: #4ecdc4 !important; color: #1a1a2e !important; } '; } function removeDarkMode() { var style = document.getElementById('universal-dark-mode-style'); if (style) style.remove(); } // Toggle with Alt+Shift+D document.addEventListener('keydown', function(e) { if (e.altKey && e.shiftKey && e.key === 'D') { e.preventDefault(); enabled = !enabled; if (enabled) { applyDarkMode(); console.log('[Universal Dark Mode] Enabled'); } else { removeDarkMode(); console.log('[Universal Dark Mode] Disabled'); } } }); // Apply on load applyDarkMode(); // Re-apply on dynamic content var observer = new MutationObserver(function(mutations) { if (enabled && !document.getElementById('universal-dark-mode-style')) { applyDarkMode(); } }); observer.observe(document.head, { childList: true }); console.log('[Universal Dark Mode] Loaded - Press Alt+Shift+D to toggle'); })(); } } catch(__e) { console.warn('[Userscript:Universal Dark Mode]', __e); } })(); })(); GitHub - Tejas-Raj01/DealLens · GitHub
Skip to content

Repository files navigation

DealLens — AI Investment Research & Due-Diligence Workflow Engine

DealLens CI PipelineLive Backend (Render)Python 3.11FastAPIPostgreSQL pgvectorNext.js 14

DealLens is a production-grade, asynchronous AI backend system designed for automated corporate investment due-diligence and financial document research. It transforms unstructured filings (Annual Reports, 10-Ks, 10-Qs, Investor Presentation Decks) into a structured vector + relational knowledge graph, executing deterministic multi-step due-diligence workflows with strict page-level citation provenance.


🌐 Live Production Deployments


Architecture Diagram

 +----------------------------------+
| Next.js 14 UI |
| (Docs, Workflows, Citations) |
+-----------------+----------------+
| HTTP / REST API
v
+----------------------------------+
| FastAPI API Gateway |
| (Validation, Auth, Middleware) |
+-----------------+----------------+
|
+-----------------------------------------+-----------------------------------------+
| | |
v v v
+-------------------+ +--------------------+ +--------------------+
| PostgreSQL 16 | | Redis 7 Broker | | MinIO / S3 Storage |
| + pgvector | | & Result Store | | (Raw Document PDF |
| (Docs, Chunks, | +----------+---------+ | Artifacts) |
| Workflows, | | +--------------------+
| Citations, Logs) | v
+-------------------+ +--------------------+
| Celery Worker |
| (Async Ingestion, |
| Workflow Engine, |
| RAG & Citations) |
+--------------------+

Executive Overview & Problem Statement

Investment analysts, private equity deal teams, and venture capitalists spend hundreds of hours manually cross-referencing multi-page corporate filings (10-K annual reports, audited financial statements, investor presentation pitch decks).

Generic RAG applications and simple chat interfaces suffer from three critical production flaws in financial engineering:

  1. Hallucinations & Groundless Claims: Models state figures like "Revenue grew 25%" without pointing to an audited line item or page.
  2. Page-Number & Provenance Loss: Traditional text splitters strip out document page numbers, leaving analysts unable to inspect the original PDF page.
  3. Fragile "Magic Autonomous Agents": Unpredictable LLM agent loops execute arbitrary steps, leading to infinite loops, high costs, and unexplainable failures.

How DealLens Solves This

  • Page-Aware Ingestion: PDF layout parsing preserves 1-indexed page boundaries for every extracted text chunk.
  • Hybrid RAG Pipeline: Integrates Postgres tsvector keyword search with pgvector Cosine Distance search via Reciprocal Rank Fusion (RRF).
  • Deterministic DAG Workflow Engine: An explicit 7-step state machine orchestrates due diligence with granular step input/output logging and automated retries.
  • Strict Provenance Guardrails: Every claim in generated investment reports is passed through a Citation Verifier to ensure zero ungrounded statements.
  • RAG & Workflow Evaluation Framework: Built-in benchmark harness (eval/evaluate.py) tracking Context Recall @ K, Citation Precision, and Latency.

Explicit 7-Step Due-Diligence Workflow DAG

Unlike non-deterministic LLM agents, DealLens uses an explicit, observable state machine:

Workflow Trigger
│
├── [Step 1] Document Validation (MIME check, SHA256 deduplication, PROCESSED state check)
├── [Step 2] Company Extraction (Entity profile, sector, US GAAP/IFRS reporting standard)
├── [Step 3] Financial Performance Analysis (Revenue, Gross Margins, EBITDA, Debt with page citations)
├── [Step 4] Risk Analysis (Regulatory, Market, Operational risk categorization)
├── [Step 5] Target Evidence Retrieval (Target vector + keyword queries for risk mitigations)
├── [Step 6] Cross-Document Claim Verification (Cross-referencing Deck claims vs 10-K audited reality)
└── [Step 7] Due Diligence Report Generation (Synthesizing memo with embedded page provenance citations)

Core Technology Stack

LayerTechnologyPrimary Rationale
Backend APIPython 3.11, FastAPI, Pydantic v2Async I/O concurrency, auto OpenAPI schema docs, strict input validation
DatabasePostgreSQL 16 + pgvectorCombined relational + vector storage, ACID transactions, HNSW vector indexing
Async Task QueueCelery 5 + Redis 7Decouples heavy PDF parsing, embedding generation, and multi-step LLM workflows
StorageMinIO / AWS S3S3-compatible raw PDF document store with SHA256 deduplication
RAG & Hybrid Searchpgvector Cosine + Postgres tsvector + RRFHybrid dense + sparse search without external vector DB operational overhead
ObservabilityStructlog JSON + Telemetry MiddlewareCorrelation IDs (X-Request-ID), request duration, and step-level audit logs
Testing & EvaluationPytest, Benchmark Eval FrameworkAutomated test coverage and quantitative evaluation (Recall @ K, Citation Precision)
Frontend UINext.js 14, React, Tailwind CSSSleek corporate dark-mode dashboard with side-by-side evidence inspector

Key System Engineering Trade-Offs

1. Why PostgreSQL + pgvector over a standalone vector database (Pinecone / Weaviate / Qdrant)?

  • Transactional Consistency: Documents, chunks, workflow runs, and citation records reside in a single relational DB. Deleting a document atomically cascades to delete its embeddings and citations.
  • Hybrid Relational Filtering: Allows combined SQL queries filtering by relational metadata (document_id, company_name, page_number) alongside vector similarity in a single query execution plan with HNSW indexing.
  • Simplified Infrastructure: Reduces operational complexity for deployment and local development.

2. Why an explicit deterministic workflow DAG over an autonomous agent framework (AutoGPT / CrewAI)?

  • Auditability & Observability: Enterprise due diligence requires predictable execution. Analysts must know exactly which step failed and inspect its inputs and outputs.
  • Cost & Latency Control: Prevents infinite loops or multi-turn agent hallucination queries.
  • Fault Tolerance & Retries: Individual steps can be retried independently without re-executing completed upstream steps.

VectorShift Backend Engineer Interview Q&A Study Guide

Q1: How does DealLens guarantee strict document provenance and prevent hallucinations?

Answer: When parsing PDFs, PDFParser maintains 1-indexed page boundaries for every extracted text block. Chunks saved to document_chunks retain their source page_number and document_id. Generated answer claims pass through a CitationVerifier guardrail that performs string matching and number-precision checks against source chunks. Unverified or contradicted claims are flagged before report synthesis.

Q2: Why use Reciprocal Rank Fusion (RRF) for hybrid retrieval?

Answer: Dense vector search (pgvector) excels at semantic search, while sparse keyword search (tsvector) excels at exact financial numbers, ticker symbols, and specific terms. RRF merges both ranked lists using $RRF(d) = \sum \frac{1}{k + r(d)}$, scoring documents consistently without needing to normalize raw cosine distances against BM25 scores.

Q3: How are long-running document ingestion and workflow execution handled?

Answer: Document uploads return a 202 Accepted response immediately with a status of PENDING. Processing is offloaded asynchronously to Celery background workers backed by Redis. Clients poll GET /api/v1/documents/{id} or GET /api/v1/workflows/{id} to track execution state without holding open HTTP connections.


Local Setup & Quickstart Guide

Prerequisites

  • Docker & Docker Compose
  • Python 3.11+

Running via Docker Compose

# Clone repository
git clone https://github.com/Tejas-Raj01/DealLens.git
cd DealLens
# Start PostgreSQL (pgvector), Redis, MinIO, FastAPI Backend, and Celery Worker
docker-compose up --build -d

Running Backend Locally (Development)

cd backend
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
# Run migrations
alembic upgrade head
# Run FastAPI API Server
uvicorn app.main:app --reload --port 8000

Running Evaluation Benchmark Script

python -m eval.evaluate

Running Test Suite

pytest -v backend/tests

License & Author

Developed by Tejas Raj as a production-grade portfolio project optimized for the VectorShift Backend Engineer — India role.

About

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages