At Digimark, I built distributed PySpark pipelines on AWS EMR for real-time automotive telemetry and developed the RAG data layer behind an enterprise chatbot. I added incremental vector indexing, embedding lifecycle management, and retrieval quality monitoring to keep answers current.
I also developed a transactional-message classification service that reached 99% accuracy and reduced manual review by 30%. For Salesforce leads, I built an identity-matching and deduplication pipeline that reduced duplicate records by roughly 25%.
Earlier, at RaftLabs, I built ETL and search pipelines across PostgreSQL, Elasticsearch, and vector indexes. At instinctools, I worked on a real-time and batch data platform handling terabytes of data and 1B+ records; I began in data engineering at TEAM International.
