AI engineer building agent infrastructure and RAG systems that run in production, not in a notebook.
Remote (IST) · Open to AI engineering roles at early-stage AI-native startups
Portfolio · GitHub · LinkedIn · Resume · naman@namangupta.dev
- Building AI tooling and backend systems at Oldowan Innovations
- Contracting on retrieval infrastructure at Infradock.ai
AgentMesh — live
LangGraph-native multi-agent orchestration with consequence-weighted human-in-the-loop gates and typed state handoffs between agents. Cut failed handoffs by 40% versus an ungated pipeline.
LangGraph FastAPI Next.js WebSockets
MCPHub — live
Observability for MCP servers: P50/P95/P99 latency tracking, schema drift detection, a JSON-RPC protocol inspector, a CI/CD CLI, and an npm package (@naman_411/mcphub-agent) for automated health checks.
Next.js Drizzle ORM Neon NextAuth
RAG pipeline rebuild — Infradock.ai
Took a production retrieval system from 55% to 95%+ precision by replacing naive chunking with clause-level chunking and cross-encoder reranking. Runs 500+ queries/day.
Python pgvector FastAPI
Python, TypeScript · LangGraph, LangChain · FastAPI, Node.js · Next.js, React · Postgres, pgvector, Qdrant · Docker, Redis, WebSockets
- NASA Space Apps Challenge 2025 — Regional Winner, Global Top 1%
- Smart India Hackathon 2024 — Grand Finalist (top 5% of 30,000+ teams)
B.Tech CSE (AI/ML), OIST Bhopal — 2026



