Li Yu

I've built and scaled production LLM applications, AI agents, and machine learning platforms at Meta for enterprise search, automation, and intelligent workflows. My work spans RAG systems, multi-agent workflows, model optimization, evaluation frameworks, and production AI infrastructure.

I designed RAG-based systems with Python, SQL, LangChain, Pinecone, and FAISS, and built agent workflows using LangGraph, LangChain Agents, and MCP. I also developed fine-tuning pipelines for Llama-based models and integrated OpenAI GPT, Anthropic Claude, and open-source Llama models to balance quality, latency, and cost.

Previously, I led research, data science, experimentation, and backend engineering work at Meta and Facebook, translating behavioral and conversational data into product and business decisions. I presented recommendations to senior stakeholders, mentored junior teammates, and supported product strategy through rigorous analysis and financial evaluation.

Earlier, I built visualization and machine learning tools at Virtualitics, game mechanics at Infinity Ward, and network deployment tools at NASA Jet Propulsion Laboratory. I bring a product-minded engineering approach to turning advanced AI capabilities into reliable systems with measurable business impact.