At Rezolve Ai, I engineered a sub-agent orchestration system for hotel bookings and concierge services, with Agent-to-Agent handoffs. I built an orchestrator with dynamic query routing that reduced hallucinations by 40% and improved intent accuracy for complex hotel operations.
I integrated real-time voice capabilities with Pipecat and LiveKit, and deployed a vector database for policy-based retrieval-augmented generation. I also established long-term memory persistence with PostgreSQL and MongoDB, using Langfuse for system tracing and performance monitoring.
On my FlixChoice project, I developed a movie recommendation engine combining content-based and collaborative filtering across 5,000+ movies. I built and deployed an interactive Flask application, achieving 85%+ recommendation accuracy.

