Yuvraj Singh

At Visa, I’ve been architecting an AI Agent Deployment & Lifecycle Management platform that replaces manual onboarding with an asynchronous state-machine workflow engine. I designed an end-to-end deployment pipeline for production AI agents and built a dual-environment abstraction to decouple QA from production infrastructure.

I also worked on resilient execution for an event-driven task dispatch system, diagnosing production concurrency failures and redesigning it with resilient polling, circuit breakers, and orphan recovery. On top of that, I built core infrastructure for an AI UI/UX Generation platform that turns Figma designs, images, or text into production-ready UI components and raises pull requests.

For repo onboarding, I created a heuristic repository analysis engine that detects monorepos, frameworks, build pipelines, and preview configurations without LLM inference. I designed a LangGraph feedback agent that supports iterative refinement without restarting execution, enabling multi-turn code generation workflows across five engineering teams.

Earlier, as a SWE Intern at Congruent Solutions, I built a Server-Driven UI SDK with React, TypeScript, and XState so backend-controlled interfaces could evolve across workflows without frontend redeployments.