Rajnish Khatri

About Rajnish Khatri

I make AI systems measurable, reliable, and trustworthy at enterprise scale. Most…

Experience

1. Principal Consultant

   Infosys

   Oct 2019 — Present · Chicago, IL, US

   As Principal Consultant for a major financial institution\'s dispute resolution systems, I architected a comprehensive multi-agent workflow using LangGraph integrating seven specialized AI agents (Intake, Analysis, Evidence, Fraud, Decision, Compliance, Escalation) with enterprise authentication. I established production-grade evaluation frameworks using LangSmith for automated performance assessment across foundation models (Claude, GPT, Llama), incorporating LLM-as-Judge systems and statistical metrics. Through DSPy framework implementation for advanced prompt optimization and declarative programming with ChainOfThought, ReAct, and ProgramOfThought modules, I achieved 30% performance improvement. The architecture features a Model Context Protocol server ecosystem with Lambda-based services providing 20+ enterprise tools, production RAG systems using AWS Bedrock and OpenSearch, and comprehensive evaluation pipelines assessing retrieval quality and answer accuracy. I implemented enterprise security with SAML SSO, JWT validation, and IAM controls, automated testing integrated with CI/CD pipelines, and comprehensive monitoring using CloudWatch and X-Ray for real-time performance insights at enterprise scale.