Exist Software Labs, Inc.
Data Engineer
AI Systems & Enablement
- Lead AI training at Exist and help teams use generative AI in delivery work.
- Built a production RAG pipeline handling more than 1,000 queries daily with LangGraph, AWS Bedrock, and PostgreSQL pgvector.
- Combined BM25 keyword search with vector similarity for hybrid retrieval.
- Built workflows for intent classification, query analysis, HyDE expansion, and semantic ranking.
- Reached 84% accuracy across 49 QA test cases through evaluation and prompt iteration.
Data Engineering & Analytics
- Automated daily regulatory and compliance reporting, cutting manual processing from 4 hours to 15 minutes.
- Consolidated agency and bancassurance data, processing more than 500,000 transactions daily.
- Optimized SQL models and Power BI datasets, cutting report load time from 5 minutes to 19 seconds.
- Refactored preprocessing with Dask, cutting runtime by 90%.