At FIS Global Solutions, I built and contributed to AI Empower, a conversational AI platform for querying financial surveillance data. I implemented its NL2SQL pipeline and RAG retrieval layer, improving reported SQL accuracy from 60% to 85%.
I also developed a multi-agent document intelligence platform for analyzing BRDs and contracts, and an internal MCP server that exposed reusable enterprise capabilities. For the MCP server, asynchronous retrieval reduced multi-source latency by about 40%.
Across other projects, I developed a RAG recommendation engine that achieved 92% accuracy on a test dataset and automated six delivery workflows with more than 90% successful execution and no human in the loop. My work includes agentic AI, cloud-native applications, and enterprise software.
