About This Project
Lead developer of the core AI agent system and major contributor to the React + Rust agent-orchestrator platform for Zemly.AI — a 24/7 property management call-handling system that autonomously answers, triages, and escalates tenant and landlord calls based on configurable agency policies. Led a team of 4 engineers on the core agent side, built in Python with Twilio for telephony, Azure for cloud infrastructure, OpenAI for conversational intelligence, and MongoDB for persistent state and call records. Beyond the agent layer, contributed substantially to the orchestrator platform — working across the React frontend and Rust backend to ensure end-to-end system coherence. Made key architectural decisions spanning both the core agent and orchestrator, directly interfaced with stakeholders, and coordinated cross-team alignment between the two systems. Additionally built an automated evaluation platform (Python server + Svelte frontend) that uses an evaluator agent to systematically test and benchmark the core agent's performance — enabling data-driven iteration on agent quality.