Koduri Mohan

I’m an AI Engineer focused on production-ready agentic AI, RAG, and secure LLM workflows. I build grounded, auditable systems with deterministic controls and permission-scoped integrations—especially where regulated enterprise requirements matter.

At Mindgate Solutions, I architected and shipped a production LangGraph agent across ~140 Python modules and 530 automated tests, keeping orchestration separate from independently tested business logic. I implemented grounded RAG and anti-hallucination flows using a grounding gate, deterministic navigation resolution, and permission-scoped access to 100+ banking API endpoints.

I also design safe multi-step tool-using workflows for payment initiation, OTP, entitlements, and maker/checker approvals while ensuring financial write paths remain deterministic and auditable. I added structured outputs, approval gates, idempotency, rate limiting, PII-safe logging, audit trails, and observable execution to strengthen trust and reliability.

Earlier, I built Java/Spring Boot payment services and secure REST APIs with maker-checker controls and audit trails, improving cross-currency payment latency by 35% with zero post-release critical defects. I’m equally comfortable with evaluation and observability—running live-backend evaluations over 83 banking FAQs to drive RAG pipeline rewrites.