Selected projects demonstrating applied AI systems, from model training to decision support.
Classifies health insurance customer queries into 21 intent categories for routing and automation. Trained using privacy-preserving, locally hosted LLM-based data augmentation.
Extracts structured tables and text from PDFs using vision-language models, with automatic model routing that reduces processing costs and optimizes accuracy.
Implements the Model Context Protocol to expose multiple external APIs as agent-callable tools. Demonstrates clean separation between tool execution and LLM reasoning using CLI and Streamlit clients.
Analyzes household profile and expected healthcare usage to recommend the best-fit medical plan from employer options.
