SOURABH MISHRA

About

Backend/AI engineer who has built a rule-service platform on AWS and evolved it into an agentic RAG system using LangChain, ReAct, tool calling, and LLM-as-judge review. Notable for combining heavy AI-assisted development with production safeguards like manual CR, CloudWatch monitoring, fallback strategies, benchmark testing, and user-feedback-driven model improvement.

Key Strengths

• Built AWS-based rule execution backend with Step Functions and Lambda orchestration
• Hands-on experience shipping AI-assisted code to production
• Implemented Redshift concurrency control with semaphore locking
• Designed LLM review pipeline using LLM-as-judge plus manual code review
• Closed production feedback loop by collecting user feedback in DynamoDB for model improvement
• Built agentic/RAG systems with LangChain, ReAct, and tool calling
• Integrated deterministic tools like Python and data APIs to reduce LLM guesswork
• Established monitoring and resilience using CloudWatch, fallback strategies, and benchmark testing
• Shows architectural judgment on code reuse and shared library design
• Evaluated multi-agent frameworks pragmatically and rejected AutoGen due to latency and debugging concerns

Reference Highlights

• very strong in AI/LLM work
• positive customer response
• end-user satisfaction
• full-stack ownership
• excellent communication skills
• keeps stakeholders in the loop
• translates BRD into HLD and LLD
• sets timelines and deadlines
• commits and delivers on time
• good prioritization
• unblocks others easily

Experience

Education

Skills

Languages