Ubaid Ur Rehman

I’m an AI Engineer & Full-Stack Developer focused on shipping agentic LLM pipelines and production full-stack applications. I specialize in deep ML with Transformers/GNNs and in building scalable, reliable infrastructure that supports real user workflows.

As an AI Engineer Intern at Skylabs AI, I automated a 4-step manual LLM workflow into a single run using agentic pipelines with built-in retry logic, cutting a multi-step process down to one automated flow. I also expanded a training dataset from ~400 seed examples to 10,000+ synthetic pairs using LLM-based generation and 15+ rounds of prompt refinement, improving dataset quality from 55–60% to 80–85%. I led research on agent-based automation for an internal team initiative, extending pipeline architecture toward end-to-end AI application development.

Previously, as a Full Stack Developer at Saynt AI, I built an AI chat feature powered by Retrieval-Augmented Generation (RAG), exposing REST APIs via Django with a TypeScript/Next.js frontend for context-grounded Q&A. I deployed backend infrastructure (PostgreSQL, Docker) on AWS EC2 with S3-based document chunk storage, optimized frontend rendering using a centralized Redux Toolkit store, and set up GitHub Actions CI/CD to run deployment tasks and service health tests before changes reached production.