Data Scientist / ML Engineer building systems that solve real, human-centered problems — and shipping most of it as open source.
- 🤖 Agent orchestration — building multi-agent systems that coordinate LLM-driven agents, tools, and workflows to tackle tasks a single model can't handle alone.
- 🧠 Large language models, up close — fine-tuning LLMs locally with PyTorch and TensorFlow, and building transformer-based models (including GPT-style architectures) end to end. I'd rather understand model internals than treat them as black boxes.
- 🧪 LLM evaluation & reliability — working with evaluation frameworks (metrics like G-Eval, faithfulness, and RAG/agentic checks) and contributing fixes upstream to open-source LLM tooling.
- 🛠️ From idea to working product — turning research ideas into things people can actually run.
- 🌐 Language-agnostic — I don't box myself into a stack; give me any language, even one you invented, and I'll build in it.
LLMs · Agent Orchestration · RAG · LLM Evals
- 🌍 Portfolio — namankhurpia.github.io
- ✉️ Email — namankhurpia2@gmail.com