At Symphony Fintech, I automated AWS provisioning for distributed trading agents behind a FastAPI and Boto3 call. I also tracked provisioned resources in MySQL to make infrastructure state recoverable and reusable.
I ran an async Python Model Context Protocol server that exposes internal trading tools to external LLM clients over Streamable HTTP. OAuth 2.0 delegated access secures the server, and its modular routing supports adding tools.
In my projects, I built a voice banking agent with a WebSocket voice pipeline and specialist agents, and an Arxiv chatbot that indexed more than 234,000 papers for semantic search. I also developed a finance controller with automated reconciliation that achieved 100% precision across 159 automatic matches on synthetic test datasets.

