Intern, Data Analytics

Authored a white paper on multi-agent AI architecture for regulated environments and built a DAG visualization tool.

What I Did

I wrote an internal white paper on multi-agent AI frameworks using Model Context Protocol (MCP) for regulated banking environments. The research focused on interpretability methods for LLM-driven automation and architectural patterns for logging decision traces. I also built FastAPI endpoints to automate data integrity checks, and created a JavaScript interface to visualize data lineage as a Directed Acyclic Graph (DAG).

Impact

The DAG visualization tool was adopted by 10+ team members and reduced time spent tracing data dependencies by about 50%. The white paper became a reference document for future GenAI initiatives. The FastAPI endpoints automated previously manual integrity checks.

What I Learned

I learned to design RESTful APIs with proper error handling and validation using FastAPI. Building the DAG visualization required understanding graph data structures and topological ordering for dependency resolution. The MCP research taught me about agent orchestration patterns and how to structure multi-agent systems for auditability.

Key Highlights

• Authored an internal white paper defining a multi-agent AI framework (using MCP) for regulated financial data environments, researching methods to mitigate LLM 'black box' risks and ensure interpretability across multiple automated workflows.

• Developed RESTful API endpoints using FastAPI to expose automated data integrity protocols and built an interactive JavaScript interface adopted by 10+ team members to visualize Directed Acyclic Graph lineage dependencies, reducing overhead by 50%.