Developed ChatDKU.com, a campus-wide agentic Retrieval-Augmented Generation (RAG) chatbot that serves the daily queries of 2,000 students and faculty. Secured $100,000 in university funding to support a locally hosted LLM inference server. Since becoming project manager in December 2025, led a 12-student team by implementing Agile workflows and overseeing stand-ups and sprint planning. Also designed the core agent logic and tool orchestration system, enabling the LLM to dynamically select, invoke, and chain specialized tools for complex multi-step reasoning, while building ingestion pipelines with LlamaIndex and LangChain to process and embed diverse documents into a Chroma vector database with Redis supporting keyword search.

Taught Reinforcement Learning concepts such as MDP, Value/Policy Iterations, Monte-Carlo, TD learning, and Q learning. Designed homeworks and coding projects for students to practically learn heavy math concepts in bite-sized code.