sakshi User

At ZS Associates, I build and ship enterprise GenAI systems for Fortune 500 pharma clients, taking products from architecture and client UAT through live deployment.

I designed a modular multi-agent sales enablement platform with DAG orchestration, planner-executor workflows, hybrid retrieval, tool integration, memory, and observability. Its reusable Python runtime reduced per-use-case integration time from more than three weeks to under one week.

I also shipped a Next Best Action engine that lifted content engagement by 20% across 200+ sales reps, and built a multilingual pre-call planning pipeline that cut runtime from more than four hours to 15 minutes. Through structured UAT across three markets, I helped drive 80% adoption within two weeks with near-zero post-launch escalations.

Outside client work, I build evaluation and retrieval systems including RAGBench, MemoryOS, and a multi-stage LLM ranking pipeline that improved NDCG@10 from 0.367 to 0.495. I care about making LLM systems measurable, grounded, reliable, and useful in production.