At Netsmart Technologies, I designed and built a production Generative AI document intelligence platform using Amazon Bedrock and Claude Sonnet. It processes more than 1,000 documents per day and eliminated 90% of manual data entry in a core customer workflow.
I architected its LLM workflow with batching, parallel execution, retries, and asynchronous processing, reducing average processing time from 60 seconds to 25 seconds. Confidence scoring, source attribution, validation, and conflict detection surface uncertain outputs for human review.
I also develop full-stack product features with Java, Spring Boot, Angular, TypeScript, REST APIs, and AWS, taking solutions from technical design through production. I lead technical design and code reviews, mentor engineers, and improve production reliability and observability.
Before this, I designed an event-driven messaging system at Netsmart Technologies that reduced failed message deliveries by 80%. I also built CI/CD and deployment workflows that reduced deployment latency by 70%, and contributed to a single-page application migration during my internship at Intouch Group.
