Shashwat Negi

About

Data Scientist (2–3 years) at ZS Associates who has built and productionized agentic LLM systems, including a LangGraph-based multi-LLM prompt-optimization pipeline for entity extraction deployed as a Spring Boot microservice via Jenkins. Also built an Insightmate.ai chatbot and improved its RAG accuracy by diagnosing vector retrieval issues and implementing HyDE query expansion, while partnering with sales and pharma stakeholders to drive adoption (e.g., Zimmer Biomet platform migration into a multi-year partnership).

Experience

Education

Key Strengths

• Took LLM agentic prototype to production with evaluation on 1,000+ annotated documents
• Designed multi-LLM prompt-optimization pipeline (Gemini generator + GPT-4 judge/refiner)
• Integrated LLM workflow into Java Spring Boot microservice and deployed via Jenkins CI/CD
• Implemented production monitoring for performance and cost; achieved 99% uptime
• Cost optimization for LLM usage (few-shot, token reduction, output limits, caching)
• Real-time troubleshooting of LLM/RAG issues using logs and targeted fixes
• Improved RAG relevance by implementing HyDE query expansion to address broad embeddings
• Strong developer-facing enablement: live code, architecture diagrams, hands-on demo environments
• Effective sales/CS partnership: technical feasibility assessments and support for client adoption