Urwah Ejaz

AI-driven Full Stack Engineer with 1.5 years of intensive, production-focused experience owning systems end-to-end from React/Node.js frontends and FastAPI/Flask backends through to cloud deployment and LLM-integrated features. Has independently built and shipped full-stack platforms (dashboards, CRMs, RAG-based products) used in live production by real teams, not prototypes. Builds with AI as a core part of the engineering workflow rather than a bolt-on, and is comfortable owning scalable architecture and product decisions within small, fast-moving teams.

AI Engineer at VisionRD, CEMTECH, NSTP – NUST (2025-06 – 2026-04)

• Full-stack ownership, end-to-end: Independently built and shipped a full-stack CRM platform for Haval automotive dealerships (backend and frontend) that surfaced live financial data from a 200+ GB MS-SQL database as real-time dashboards and analytics, used in production by 20+ dealerships.
• Scalable architecture & AI agents: Designed the CRM's analytics layer using agent orchestration, MCP (Model Context Protocol), and a vector database to turn raw dealership data into actionable insights owning the architecture from database to UI.
• Generative AI & LLM integration: Engineered Intelligence Hub, a full-stack analytics platform combining deep learning-based time series forecasting with prompt engineering and LLM-generated insights over multi-source structured and unstructured data, served via Flask REST APIs.
• Computer vision at the edge: Built Factory Copilot, an edge-deployed wearable AI system (dual ResNet backbones, Smart Gated Fusion, Memory-TCN heads) for real-time assembly-error detection, delivering a 35% improvement in factory efficiency in live production deployment.
• Internal tooling: Built a custom data annotation tool (backend and frontend) from scratch for annotating raw factory video data, used internally by the team.
• MLOps & deployment: Implemented pipelines using Git, GitHub Actions (CI/CD), and DVC for automated training, versioning, and reproducible experimentation, containerizing and serving systems with Docker on AWS.

AI Intern – Generative AI & NLP at AIM Lab, FAST-NUCES (2024-06 – 2025-06)

• AI Interviewer Bot: Led development of an LLM-based conversational agent, integrating NLP pipelines and prompt engineering to automate candidate assessment and multi-turn dialogue.
• Applied research collaboration: Collaborated with academic researchers to prototype, evaluate, and QA-test AI-driven assessment systems for real-world deployment scenarios.

Bachelor of Science in Artificial Intelligence – FAST-NUCES, National University of Computer and Emerging Sciences (2026)