KI / ML · Robotik · ROS2 · Gazebo · Computer Vision
Raja Hammad Naseer
KI- & Robotik-Systeme — von der Wahrnehmung bis ins Produkt.
M.Sc. Computer & Systems Engineering student building AI/ML systems and ROS2-based autonomous-vehicle architectures with real-time perception — and integrating them into full-stack products that actually ship.
M.Sc. Computer & Systems Engineering student with industry experience in AI/ML systems, software engineering, and technical project collaboration. Experienced in developing and integrating complex software systems — including ROS2-based autonomous-vehicle architectures and real-time perception pipelines. Strong analytical mindset with an interest in systems engineering, structured problem solving, and interdisciplinary technical coordination.
0+
years full-time industry experience
0
production systems shipped & live
0
web apps with zero downtime
C1 · B1
English · German Urdu native
02
Werdegang / Experience
Von Idee zu Produkt
2023 — 2026
2026 — heute · Ilmenau
Machine Learning Intern
Fraunhofer IOSB (Kognitive Energiesysteme)
Extending Amazon's Chronos foundation model for energy time-series forecasting: load, generation, electricity prices and weather data.
Profiling the existing implementation for accuracy, robustness and runtime against public energy-market datasets.
Designing and implementing extensions to the model architecture, adapting foundation-model methods to how energy markets behave.
2024 — 2025 · Pakistan
Full-Stack Developer
Jataq Technologies
Shipped three production web apps here, building the full Node.js/React stack and running the Docker CI/CD pipelines myself, from first commit to release.
Support was buried in repetitive questions, so I built a RAG pipeline to answer them, tuned the prompt chain, and wired it into the existing site over REST.
When the team needed a direction for its AI tooling, I benchmarked the main LLM frameworks, presented what I found, and we put the winner into the production stack.
2023 — 2024 · Pakistan
Artificial Intelligence Engineer
VisionTech360
EMACS was my main build: a full access-control platform running live across several camera sites. React dashboard, Node/Express API, MongoDB, all in Docker.
Underneath it sits a concurrent backend I designed to handle several video feeds at once, tracking whitelist state in real time and timestamping every event.
I took it the whole way, proof of concept to production, running the YOLOv8/ONNX inference as its own service behind a repeatable Docker release.
03
Ausbildung & Forschung / Education
In der Forschung verwurzelt
control · CV
Okt 2025 — heute · Deutschland
M.Sc. Computer & Systems Engineering
Technische Universität Ilmenau
Note 1,3 · Research Seminar
TEC Cooling Control
Modelled a TEC system (PT2 + dead time), identified parameters via open-loop step test, and tuned a PID controller with anti-windup on Arduino. Rise time 34 min, steady-state error <0.3 °C.
Laufend · SS2026 Group Study
Modular ROS2 Autonomy
Designing a modular autonomous-driving architecture with defined subsystem interfaces for perception, control and safety — coordinating multi-agent interaction in Gazebo Sim.
Note 1,3 · Research Skills
UAV Multi-Object Tracking
Built a real-time tracking pipeline and benchmarked ByteTrack vs. Norfair on MOTA, ID-switch rate and FPS across simulated UAV flight maneuvers.
2019 — 2023 · Pakistan
B.Sc. Computer Science
SZABIST
04
Ausgewählte Arbeit / Selected Work
Dinge, die live gingen
6 systems
01
▸ AI · Retrieval / Backend
Air-Gapped Semantic Search Engine
Three microservices — an ONNX embedding service (~3× faster on CPU), HNSW vector search returning the top-30 in ~10 ms, and a cross-encoder reranker for top-10 precision. Fully local, no external APIs.
A React.js dashboard + Node.js/Express REST API + MongoDB, deployed via Docker across multiple live camera installations. The backend integrates a YOLOv8/ONNX inference service for real-time face recognition and whitelist state management.
ReactNode.jsExpressMongoDBDockerYOLOv8ONNX
Live · multi-sitePrivate · NDA
03
▸ Fullstack · Computer Vision
WorkAI — Industrial AI Monitoring
A full-stack monitoring platform: a React.js real-time dashboard, Node.js REST APIs and MongoDB event logging, all containerised with Docker. Pose-estimation and face-recognition pipelines are exposed as backend services.
A full-stack student marketplace built from scratch: user profiles, listings, search, filtering, an admin panel and a responsive UI — validated through user interviews with TU Ilmenau students.
Upload multiple CVs and paste a job link or description — CoverCare analyses your experience against the role and generates a tailored cover letter, then exports it as a polished PDF. Dual AI engine: Gemini 2.5 or local Llama 3.2 via Ollama.
An LLM agent with dynamic tool use: live web-search grounding, chain-of-thought reasoning, and structured argument / counter-argument output with citation tracking.
