On REA, I engineered Model Context Protocol tool-calling interfaces for AI agents, resolving function-calling dialect incompatibilities across Google Gemini and LLM client SDKs.
I improved tool execution reliability by normalizing JSON Schema constraints while preserving prompt guidance. I also maintained server-side runtime validation and wrote automated tests against live model constraints.
I built Dev-Mem, a local-first memory engine for AI coding agents, with an evidence-linked state graph to reduce cross-session context loss. Its GraphRAG pipeline and regression interception system were evaluated across agents.
I also built Arc, a competitive programming analytics platform, and TrailPlan, a travel planner that uses Google Gemini to generate budget-optimized itineraries. I’m studying Computer Science at Pimpri Chinchwad College of Engineering (PCCOE).
