- 🤖 Building production Agentic AI systems — LangGraph agents, RAG pipelines, pgvector, multi-agent orchestration
- 💻 3.5+ years shipping full-stack web applications end-to-end — architecture → development → AWS deployment
- 🏆 Meta Hacker Cup Round 1 Qualifier · 2000+ competitive programming problems solved
- 🌍 Based in Ambikapur, Chhattisgarh, India
| Tool | Purpose |
|---|---|
| LangGraph | Stateful multi-agent orchestration graphs |
| LangChain | LLM application framework, chains & tool calling |
| RAG Pipelines | Retrieval-Augmented Generation with vector DBs |
| pgvector / FAISS | Vector search — hybrid keyword + semantic |
| LangSmith | Tracing, evaluation & observability |
| MCP | Model Context Protocol tool integrations |
| OpenAI / Claude API | LLM integrations, structured outputs & tool use |
Objective-driven research agent: a LangGraph pipeline (planner → researcher → analyst → quality-check loop → report) where the user's goal steers every node prompt — so an acquisition brief and a partnership brief come out materially different. Streams node-by-node via SSE, Postgres-checkpointed state, grounded follow-up chat.
Tech: LangGraph · FastAPI · React · PostgreSQL · Tavily · Docker
Semantic anime recommender: offline FAISS index → embed query at request time → MongoDB metadata filters → LLM re-rank to ~12 picks, each with a one-line "why". AniList OAuth personalisation from watch history, LangGraph chat agent, weekly precision@k eval in CI — all on free-tier infrastructure.
Tech: Next.js · FastAPI · FAISS · MongoDB · LangGraph · LangSmith
🏘️ Neki — AI Content Platform(Production)
AI content-creation platform for nonprofits. Production LangGraph agent (FastAPI) for multi-org social content generation, pgvector RAG knowledge base with HNSW indexing and hybrid keyword + vector search, Meta Graph API integration, multi-tenant architecture. Sole developer for 2.5+ years.
Tech: Next.js · Node.js · Python · LangGraph · PostgreSQL · pgvector · AWS
🪄 AI Social Content Agent(Production)
Multi-org social content generation agent powering Neki. A stateless LangGraph (FastAPI) pipeline that assembles bounded, org-scoped context from a pgvector knowledge base and drafts on-brand posts, with a resilient image-generation fallback ladder (gpt-image-1 → mini → dall-e-3 → dall-e-2 → PIL) and one-click publishing to Facebook, Instagram & LinkedIn via the Meta Graph API. Migrated from a Streamlit prototype to a decoupled Next.js + TanStack Query + FastAPI stack, with Chainlit as an optional debug harness.
Tech: LangGraph · FastAPI · Next.js · TanStack Query · pgvector · Meta Graph API
- ✅ Solved 2000+ coding problems across LeetCode, Codeforces & CodeChef
- 🏆 Meta Hacker Cup Round 1 Qualifier
- ⚡ LeetCode — codeantik_099
constankit={name: "Ankit Singh",roles: ["AI Engineer","Full Stack Developer"],stack: {ai: ["LangGraph","LangChain","RAG","FAISS","pgvector","MCP"],backend: ["Node.js","FastAPI","Express","PostgreSQL","MongoDB"],frontend: ["Next.js","React","TypeScript","Tailwind CSS"],infra: ["AWS","Docker","Kubernetes","CI/CD"],},currentFocus: ["Production agentic AI systems","Multi-agent orchestration with LangGraph","RAG & hybrid vector search pipelines","Scalable full-stack SaaS architectures",],funFact: "Solved 2000+ CP problems and still thinks in graphs 🧠",};⭐ From codeantik