AI Networking Assistant with Semantic Matching

I built a networking-focused system that stored professional profiles in Pinecone vector database and enabled semantic search to match users based on natural language queries. The system could automatically initiate Twilio conference calls between matched professionals.

Implemented profile embedding where user information (skills, goals, location, industry) was converted to vector representations, enabling semantic similarity search beyond keyword matching. Built tools for the AI to extract profile information from conversations, search across stored profiles, and orchestrate multi-party calls using webhooks.

Through this project, I explored the UX challenges of agentic AI systems: when to ask for user confirmation vs. acting autonomously, how to handle search result ambiguity, and the technical infrastructure needed for real-time AI decision-making.