Jenit Jain

Created a unified data model for more than 20 fintech core systems supporting over 17 million customers. Implemented Databricks model serving and MLOps orchestration for semantic search used by finance, credit, and AML teams. Built a GraphRAG retrieval pipeline over credit policy documents using hybrid retrieval, Databricks Vector Search, and reciprocal rank fusion. Developed an LLM-as-judge evaluation framework calibrated against human ratings and architected a LangGraph multi-agent supervisor with prompt-injection safeguards, Microsoft Teams integration, and Plotly visualizations.

Designed a document-processing SaaS platform combining OCR-based text extraction with NLP models and workflows aligned to underwriters' needs, reducing deployment turnaround by 70%. Developed a key-information extraction technique that improved accuracy by 18% and built a recommender system combining collaborative and content-based filtering. Promoted reusable Docker assets for graph-based neural networks and retrieval systems, accelerating proof-of-value work by 55%, and deployed containerized inference services on AWS EKS for real-time workloads.

Implemented data-driven automation using time-series data collected from more than 10 sensors and validated improvements through A/B testing.