I built Insurellm, a RAG-based knowledge assistant using LangChain, ChromaDB, and OpenAI. I created an evaluation harness across 150 queries and used its findings to improve retrieval and answer quality.
For a deal-hunting agent system, I combined RAG with a fine-tuned pricing model to surface underpriced listings. I deployed model inference on Modal and orchestrated reasoning across OpenAI, Anthropic, Google Gemini, and local Ollama models.
I also developed a network intrusion detection pipeline that achieved a 0.90 F1-score on 1M+ records, with MLflow tracking and CI/CD deployment to AWS. I’m applying my Chemical Engineering foundation in mathematics and statistics to building and evaluating AI systems.
