AI Engineer

I build AI agents from scratch — not on top of frameworks, but from the ground up. My macOS agent has multi-agent orchestration, a Planner-Critic-Judge loop reviewing every action before it runs, a cost-tiered execution engine across seven routes for roughly a 10x token cost reduction, episodic memory with Bayesian-smoothed priors, and an LLM-as-judge eval harness wired into CI. I also built RAG pipelines with LangChain and FAISS, an OCR pipeline outputting structured JSON with per-field confidence scores, and shipped TryOwn, a live full-stack AI app on Next.js, FastAPI, and pgvector validated with about 20 real users. About a year of experience, based in Delhi/NCR, open to remote or relocating. Looking for an AI/ML engineering role at a company building real agent infrastructure — evals, reliability, orchestration, or production deployment.