Full-Stack AI Engineer
Part-time at DevNauts · Independent Freelance · Solving very specific problems on web
- Building production RAG and solving specific problems
- Full-stack work in Next.js + TypeScript for product companies and agencies
- Open source contributions in TypeScript/Next.js codebases
Some things I've dealt with building real software along my journey:
- Built JS normaliser scripts to parse messy Equifax and TransUnion credit reports for a Nigerian SaaS product. Dealt with inconsistent structure, shifting field names, and mapped them to a clean internal schema that a live dashboard could actually use.
- Implemented a centre-indexed bar chart in Chart.js 2 for a US client. Zero at centre, bars going both directions, stable at scale. No docs for it. Went through the library source and built a custom plugin.
- Built a full PDF-to-searchable-knowledge pipeline — Marker models, AI-enriched chunking, hybrid BM25 + vector search in Qdrant — because off-the-shelf solutions couldn't handle math-heavy PDFs the way I needed.
- This was pre-GPT. A client needed Easebuzz integrated into his NGO app, which had weak documentation and no tutorials at that time. I read through the raw API references, contacted the Easebuzz team directly for clarification, and reverse-engineered the auth and webhook flow. Got it working and shipped.
Stack: TypeScript, Next.js, React, Node.js, FastAPI, PostgreSQL, MongoDB, Redis, Docker, LangGraph, Qdrant, Mem0,
