Usman Khan
AI & Full-Stack Engineer
I build real-world software products — from intelligent AI systems and RAG pipelines to scalable full-stack applications designed for production use.
Work Experience
PersonalClientCare (PCC)
Full-Stack Engineer (Werkstudent)
Sep 2025 — Present- Developed user-facing features and backend automation for a production application used in real workflows.
- Designed and implemented backend APIs and data pipelines to support intelligent automation.
- Shipped scalable, production-ready improvements in a fast-paced engineering environment.
LeadLane
Founding Engineer (Werkstudent)
2025 — MVP Engagement- Built the AI-powered MVP of a B2B sales automation platform from concept to production.
- Designed and implemented RAG pipelines using vector databases and LLMs (LLaMA 3, Mistral).
- Improved outbound sales by ~70% and operational efficiency by ~80%.
- Engineered low-latency AI pipelines delivering responses in under 3 seconds.
Featured Projects
Context
A fast, RAG-first AI chat platform inspired by t3.chat, built for document and YouTube-based knowledge retrieval with a strong focus on latency and UX.
CourseVault
A full-stack platform for university students to discover, share, and rank study resources, deployed on Vercel as a complete production system.
DreamTransport
Public transport optimization system for Neumarkt i.d. OPf. to improve bus routes and stops using passenger data, punctuality analysis, and mathematical optimization algorithms.
Klebl
Hackathon-winning AI system for automated architectural blueprint analysis using YOLOv11 and Gemini, reducing manual review time by ~90%.
Voice Oasis
Hackathon-winning real-time voice enhancement and conversational memory system with speaker recognition, designed under strict latency and privacy constraints.
Offline_AI
An offline-first Android application enabling on-device LLM inference and agent workflows, focused on privacy, reliability, and low-latency execution.
MHack
A desktop AI assistant capable of browsing, code execution, and task automation through agent-based tool orchestration.
Hackathons
MITTELSTAND HACKATHON
Winner (1st Place)Klebl GmbH challenge. Built Klebl AI. Used YOLOv11 and Gemini API to analyze construction blueprints, reducing analysis time by 90%.
{Tech: Europe} AI Hackathon
3rd PlaceBuilt UdS OP (Universal Desktop Assistant). AI assistant capable of code execution, file operations, browsing, and searches.
START HACK 2025
3rd PlaceBiggest one I've ever attended. Co-developed VoiceOasis, a real-time, privacy-focused voice AI for robotics featuring speaker recognition and conversational memory.
Servus AI Summer
2nd PlaceDreamTransport: Optimization system for Neumarkt i.d. OPf. to improve bus routes using passenger data, punctuality analysis, and mathematical algorithms.
About Me
Hi, I’m Usman. I’m a Computer Science student at Universität des Saarlandes and a working AI & Full-Stack Engineer. I focus on building production-grade software — not just models or demos — but complete systems that handle real constraints like latency, privacy, and scalability. My work spans AI-powered products, real-time voice and vision systems, offline-first applications, and full-stack platforms deployed in production. I enjoy working in early-stage and fast-moving environments, where ideas quickly turn into shipped products.