Intern, Data Analytics

Built a multi-agent Q&A tool for internal teams and migrated manual SQL queries into Airflow pipelines.

What I Did

I built a multi-agent Q&A tool using LangChain and LangGraph backed by AWS Bedrock. The agents could query Oracle databases with SQL and read Confluence pages, then serve cited answers to software engineers and analysts through a FastAPI interface. I also migrated 35 SQL queries the compliance team had been running manually into scheduled Airflow pipelines pulling from Oracle.

Impact

The Q&A tool gave software engineers and analysts a single interface to get cited answers from both databases and documentation. The Airflow migration eliminated manual execution of 35 compliance queries.

What I Learned

I learned to build multi-agent systems with LangChain and LangGraph, including agent routing and tool orchestration. I gained experience with AWS Bedrock for LLM inference and Airflow for scheduling and monitoring data pipelines. Working with Oracle and SQL in a production banking environment taught me about query optimization and data governance.

Key Highlights

• Built a multi-agent Q&A tool in LangChain and LangGraph backed by Bedrock, where agents queried Oracle with SQL and read Confluence pages, serving cited answers to software engineers and analysts through FastAPI.

• Moved 35 SQL queries the compliance team ran manually into Airflow pipelines pulling from Oracle.