Osama A.Private profile: Upwork shows its earnings only to signed-in members, so we check it every few weeks, not daily. MRR here is a rough estimate (Upwork's six-month earnings ÷ months active). Last read Nov 5, 2026.

optimize and deploy deep learning models on resource-constrained hardware — from edge GPUs (NVIDIA Jetson, A100/H100) to neuromorphic chips. My work sits at the intersection of AI research and production engineering. What I deliver: — Model compression & quantization (8-bit, binary inference) that cuts latency by 30%+ without accuracy loss — End-to-end computer vision / LLM pipelines for industrial applications (YOLO, Swin-Transformer, real-time detection) — Custom CUDA kernels and GPU-optimized training pipelines — Multi-agent reinforcement learning systems for robotics and industrial automation — Research-grade work: 3 peer-reviewed publications, experience at TRUMPF, Gemesys, and Unicontrol Background: M.Sc. Automotive Software Engineering (TU Chemnitz), M.Phil. Engineering (UTM Malaysia). Former AI Engineer at Gemesys GmbH (neuromorphic hardware), Senior AI/ML Engineer at Unicontrol (automotive vision systems), and Master Thesis Researcher at TRUMPF (multi-agent RL for industrial robots). I approach every project by modeling the problem first — typically with NumPy for simulation — before building the production solution. If you need an AI system that works on real hardware under real constraints, let's talk.