Ritwik Rituparn

I’m a Delivery Engineering Team Lead focused on production-ready agentic AI systems. At Turing, I architected MCP server infrastructure for Amazon’s Agentic AI project and built multi-domain knowledge databases from scratch across 6 production domains.

I engineer tools and workflows that help autonomous LLM agents retrieve, reason over, and act on live structured data. I created 100+ MCP-compatible Python tools (CRUD, lookups, order-status, inventory APIs) and designed end-to-end agentic task flows in JSON with standardized schemas across 5+ environments.

I also drive model-alignment improvements through hands-on RLHF and fine-tuning. I executed RLHF tasks for Apple and NVIDIA across 500+ annotation tasks, built SFT notebooks in JavaScript, Python, and SQL, and reduced model cold-start errors by ~25%.

As Team Lead, I manage 15 analysts and engineers via standups, sprint planning, and skill-based allocation, while serving as a technical escalation point for MCP schema and agent failure-mode decisions. Earlier, I interned on software engineering in Java/JavaScript/MySQL, and I’ve built RAG and multi-agent systems (e.g., HR Cortex) with evaluation and observability using LangSmith.