AI engineer in Bangalore. I work on LLM agents and reasoning, and lately I've been training small models with RL to see how far I can get on free GPUs.
Right now I'm at Mea, working on LLM extraction for insurance documents: prompt tuning, digging through model outputs and Lambda logs, and helping build an agent that writes prompts from entity descriptions.
I contribute to Graphify, a tool that turns codebases into knowledge graphs. My fixes there have shipped in releases: C# call resolution through casts, Scala context bounds, edge direction in graph diffs, and a few extractor bugs.
Python, PyTorch, Hugging Face (TRL, PEFT, LoRA/QLoRA), LangChain/LangGraph, RAG and vector DBs, SQL, Docker, AWS and Azure.
LinkedIn · mp5272672@gmail.com · open to ML/AI engineering roles
