Sky

Architected a high-scalability multi-agent framework using concurrent sub-agents to parallelize complex data extraction, automated reporting, and intelligent web-scraping workflows. Developing an automated autoresearch framework to streamline and scale quantitative research, and engineered custom automation tools via Model Context Protocol (MCP) servers, Amazon Q, and advanced tool-calling to build self-service data-mapping pipelines that remove manual spreadsheet bottlenecks. Built a self-improving meta-agent simulation testbed to benchmark prompt robustification and model accuracy under distribution shifts — and served as a technical representative presenting complex AI architectures to enterprise clients and executive stakeholders.

Built a multi-agent system with an LLM-powered natural language query interface for Portfolio Analytics, combining structured and unstructured data. Engineered real-time data pipelines with quant researchers and PMs, and delivered an internal framework for dynamic code generation and execution via agent graphs — adopted across the Asset Management LOB.

Built an ML-driven vendor churn model from high-dimensional data, and an end-to-end pipeline for embedding, storage, and mapping of 3M+ menu items across companies. Improved reliability of mission-critical data by 10% via snapshot models and data-integrity fixes.

Sole programmer on a Unity / C# game prototype — designed and built the core mechanics end to end, with a bot system, procedural map generation, and round-based card mechanics, working alongside level designers and artists.