At UST, I've progressed from Software Engineer to Lead I Data Scientist, building enterprise LLM applications for financial and healthcare insurance workflows.
I designed a LangGraph-based financial report generation platform in Databricks, combining OpenAI models, few-shot examples, response voting, Python data pipelines, and PostgreSQL. The automation saved approximately 400 hours annually for fund managers.
I've also built a production-oriented healthcare insurance RAG system with LangChain and Llama 2, improving retrieval through document preprocessing, parent-child chunking, vector retrieval, and reranking. I focus on measurable correctness, grounding, traceability, and auditability through evaluation and observability tooling.
