Ken Lu

Python, TypeScript, PyTorch, Hugging Face, LangChain, vector databases (pgvector, Qdrant), Docker, FastAPI, Next.js, prompt engineering, RAG, evals, fine-tuning (LoRA / QLoRA), vLLM

Multi-agent orchestration (LangGraph, CrewAI), tool-using agents with constrained decoding, and on-device SLMs for privacy-first apps.

An open-source agent harness that turns a single prompt into a multi-step research workflow with citation tracking. Stack: TypeScript, Bun, Qdrant, gpt-4o-mini.

Technical writing, conference talks, podcast guest spots, and consulting on LLM integration, RAG design, and AI dev-tooling. Open to long-form collaborations with research teams.