Backstage interview
summary
Generative pipeline moves from POC to production: deployed on AWS and stitches LLMs, RAG and vector search with post-filtering and retry logic to handle noisy outputs; benchmarked prompts and guardrails ensure SME-level extraction quality; applied to marketing claims extraction and reduces annotation time by a great percentage in a live product; for sensitive data the stack uses open FGA for row-level authorization to enforce who sees which documents.
AI-written summary · about
related talks
30mProductionizing an unsupervised machine learning model to...
27mData storytelling in the era of generative AI
40mPreparing data for LLM training with Data Prep Kit
40mBuilding highload ML powered service
23mBuilding Production-Ready Data Pipelines for AI/ML with Lambda Architecture
29mAI-Powered Anomaly Detection in Cloud Data Pipelines
38mQuerying Live Data With LLM App
42mBuilding a Data Driven Organization