Building reliable data systems that turn operational complexity into measurable decisions.
BACKGROUND · FOCUS · HOW I WORK
I design and operate reliable data systems that transform operational data into measurable business decisions.
My background is rooted in mission-critical infrastructure and 24/7 enterprise operations, where reliability, observability, security, and incident response are part of the system — not afterthoughts. That operational foundation shapes how I design pipelines, APIs, automation, analytical products, and production services.
My core focus is Data Engineering: ETL and ELT pipelines, event-driven ingestion, data modeling, operational analytics, APIs, data quality, and observability. I also build RAG, semantic retrieval, forecasting, and local-LLM capabilities when they provide clear operational value.
AI augments how I explore, implement, test, document, and improve systems. Architecture, validation, security, and technical decisions remain grounded in software engineering principles.
Important
Current focus — Operating an enterprise helpdesk analytics platform that processes 21,000+ support tickets, combining scheduled incremental ingestion, business-hours MTTR, configurable SLA normalization, trend analysis, statistical projections, and interactive operational dashboards.
FOUR DISCIPLINES · ONE SYSTEM
Data Engineering Incremental ingestion, ETL/ELT, validation, deduplication, data modeling, analytical storage, scheduling, retries, and audit trails.
| Operational Analytics Data products for tickets, events, logs, SLAs, infrastructure health, incident signals, and business-facing operational indicators.
|
Applied AI & Retrieval RAG, hybrid search, semantic retrieval, reranking, local inference, time-series forecasting, evaluation, and grounded generation.
| Systems & Reliability Backend services, Windows tooling, secure automation, real-time agents, observability, CLIs, and performance-oriented software.
|
PRIVATE CASE STUDIES · INTENTIONALLY ANONYMIZED
Note
Client names, infrastructure identifiers, operational thresholds, proprietary data, and incident details are omitted from every case study below.
Enterprise Helpdesk Analytics Platform Operational intelligence platform that transforms support-ticket lifecycles into reliable SLA, MTTR, backlog, workload, trend, and projection indicators.
STACK Python · FastAPI · DuckDB · Polars · React 19 · TypeScript · TanStack Query |
Distributed Fleet Observability & Incident Response Multi-site observability platform for distributed Windows infrastructure, combining continuous health probes, remote inventory, event and log ingestion, persisted operational history, and real-time dashboards. The platform includes low-overhead monitoring agents, transition-based alerts, integrity-verified operational artifacts, and guarded workflows designed around allowlists, dry runs, explicit confirmation, and auditability.
STACK C# · .NET 8 · Rust · PowerShell · TypeScript · Fastify · React · SQLite · SSE · Prometheus |
Time-Series Risk Forecasting & AI Decision Support Forecasting pipeline that combines live public data, 26 years of historical time-series data, weather signals, bidirectional LSTM inference, and multi-model LLM synthesis with local inference. The system applies schema validation, retry and fallback strategies, atomic caching, dependency injection, cross-model consistency checks, and automated testing to produce traceable technical reports.
STACK Python · PyTorch · BiLSTM · CUDA · Public APIs · Local LLMs · pytest |
Enterprise Knowledge & AI Platform Internal knowledge platform with structured technical documentation, AI-assisted retrieval, protected authentication flows, role-based access control, audit trails, sensitive-data filtering, and conversational support workflows.
STACK Next.js · TypeScript · Fuse.js · OpenAI-compatible APIs · JWT · RBAC |
PUBLIC REPOSITORIES · ENGINEERING PORTFOLIO
py-rag-engine End-to-end RAG engine with PDF and Markdown ingestion, recursive and semantic chunking, PostgreSQL and pgvector persistence, hybrid retrieval, Reciprocal Rank Fusion, Cross-Encoder reranking, grounded generation, and offline quality evaluation. STACK Python · FastAPI · PostgreSQL · pgvector · SQLAlchemy · RAGAS · pytest |
Beam Secure local clipboard synchronization between Windows and Android. Supports text and image synchronization, LAN discovery, encrypted sessions, device pairing, screenshots, simulated keyboard input, and automatic reconnection — without a cloud relay. STACK C# · .NET · WPF · Kotlin · Jetpack Compose · WebSocket · X25519 · AES-256-GCM |
wallet-live Transactional investment wallet built entirely in Rust, with exact monetary values, immutable transaction history, market quote synchronization, secure session management, role-based authorization, SSR, observability, and containerized deployment. STACK Rust · Axum · PostgreSQL · SQLx · Askama · HTMX · OpenTelemetry |
Completed Go engineering labs
| Repository | Systems concepts |
|---|---|
| godistributedkv | Quorum replication, leader election, gRPC, and write-ahead logging. |
| goobservabilitystack | RED metrics, OpenTelemetry, Prometheus, Jaeger, and Grafana. |
| gocontainerruntime | Linux namespaces, cgroups, chroot, and process isolation. |
| gowasmrunner | Isolated WebAssembly execution with memory and timeout limits. |
Completed study and systems-engineering repositories, currently archived.
DATA · BACKEND · SYSTEMS · OBSERVABILITY
View the complete engineering toolkit
| Layer | Technologies |
|---|---|
| Data & Backend | Python · PostgreSQL · pgvector · DuckDB · Polars · SQLAlchemy · SQLx · FastAPI · Axum · Fastify · Next.js |
| AI, Retrieval & Forecasting | PyTorch · CUDA · BiLSTM · Cross-Encoder reranking · RAGAS · local LLMs · OpenAI-compatible APIs · Fuse.js |
| Systems & Applications | Rust · C# · .NET 8 · WPF · Kotlin · Jetpack Compose · React 19 · HTMX · Askama · PowerShell |
| Platform, Automation & Observability | Docker · OpenTelemetry · Prometheus · Grafana · Jaeger · gRPC · SSE · WebSocket · SQLite · pytest |
WHAT I OPTIMIZE FOR · AND WHY
| Principle | How it appears in my work |
|---|---|
| Correctness before dashboards | Typed schemas, validation, deduplication, deterministic business rules, and explicit data-quality checks. |
| Reliability by design | Idempotency, retries, fallbacks, atomic writes, checkpoints, recovery paths, and failure-aware workflows. |
| Observable by default | Structured logs, metrics, traces, health checks, audit trails, historical state, and actionable alerts. |
| Security with operational controls | Least privilege, encryption, allowlists, dry runs, explicit confirmation, secret isolation, and auditable actions. |
| AI with measurable value | Retrieval and models are evaluated, grounded, constrained, and applied only where they improve an actual workflow. |
AI-assisted engineering workflow
AI tools support exploration, implementation, testing, documentation, and review. They operate inside an engineering workflow built around specifications, source control, automated tests, static analysis, security review, and human validation.
LIVE METRICS · POWERED BY esousa97.com
