About
Durga Prasad Pradhan
AI Full-Stack Engineer Bengaluru, India
I build end-to-end web products and AI-powered research systems, with about two and a half years of shipping to production.
I studied engineering at NIT Rourkela, and in June 2024 joined Immigreat AI as its founding full-stack engineer. In a two-person team, where I make most of the engineering decisions, I build GreenCaard, an AI research SaaS and learning platform: the subscription AI Decision Analyzer, the course storefront and LMS, and the data pipeline behind them.
On my own time I build AI systems in the open, with their diagrams and evals published: Baatein AI, a Hinglish voice companion that remembers you with your permission, and Lexi, a legal precedent research agent.
What I build, and want to keep building: AI agents, legal-domain assistants, internal automation and data ingestion pipelines, as complete applications. I’m most interested in agentic systems, retrieval, voice AI and serverless architecture, and lately I’m exploring personal AI advisors for founders and recruiters.
How I work
- Ownership
- I take systems end to end: architecture, code, deployment, security and scaling, not one slice of them.
- Judgment
- I can explain every major decision I made and the trade-off behind it.
- Verification
- I measure AI systems with evaluations instead of trusting how they look in a demo.
Experience
Jun 2024 – present
Founding Full-Stack Engineer, Immigreat AI
Building GreenCaard end to end: the subscription AI Decision Analyzer, the course storefront and LMS, and the data pipeline behind them, from product and database to CI/CD, security and scaling. Two-person team; I make most of the engineering decisions.
Education
2020 – 2024
B.Tech, Civil Engineering
National Institute of Technology Rourkela
Skills
- Languages
- TypeScript, JavaScript, Python, SQL, HTML, CSS
- Web engineering
- Next.js, React, Node.js, FastAPI, PostgreSQL, pgvector, Supabase, REST APIs
- AI & retrieval
- LangGraph, Gemini, Anthropic Claude Haiku, Voyage AI, RAG, Pinecone, BM25, hybrid retrieval, bge-reranker, evaluation (DeepEval / G-Eval)
- Voice
- LiveKit Agents, Sarvam, OpenAI Realtime API, WebSockets, real-time speech, text-to-speech
- Cloud & DevOps
- AWS EC2/S3, Cloudflare Workers/R2, Vercel, Railway, Docker, GitHub Actions, Apache Airflow
- Quality & monitoring
- Playwright, k6, Sentry, LangSmith, Redis/Upstash
- Product
- Clerk, Stripe, Resend