Hamza Z.

Most CV freelancers can show you a model. I can show you a working multi-sensor perception system running on real data. See the demo in my Portfolio section below. I’m a perception and computer vision engineer with 10+ years of experience building 3D perception stacks and deploying them into real pipelines. My stack covers 4 cameras, LiDAR, radar, IMU and GNSS, synchronized in ROS2, with tracking, fusion and a live telemetry dashboard. What I can help with: • Multi-object tracking (ByteTrack, AB3DMOT, Kalman filtering) • Camera–LiDAR–radar fusion and calibration-based projection • 3D object detection integration (PointPillars / OpenPCDet) and LiDAR-based 3D box estimation • BEV perception (BEVFusion on nuScenes, currently in development) • ROS2 pipelines: multi-sensor sync, rosbag replay, DDS performance tuning • Python to C++ porting (Eigen, pybind11), validated against a reference implementation • Inference optimization: TensorRT / ONNX, FP16 / INT8 (3.5 to 26 FPS, ~7.4×) • Satellite and remote-sensing segmentation, OCR, industrial defect detection Recent results: → C++ port of a fusion core: ~1.78× lower latency (273.7 ms to 153.9 ms) with exact parity on detections, track IDs and counts → Distributed training across ~50 GPUs at >90% utilization → OCR accuracy improved from 85% to 98% How I work: ✔ Clear communication: you’ll always know what’s built, what’s tested and what’s left. ✔ Measured results: I report latency, FPS and accuracy, not just “it works.” ✔ Reliable delivery: I finish milestones on time and don’t give up on hard problems. What clients say: → “He was fantastic. Did not give up on challenges and completed the milestone we set successfully!” Send me your sensor setup and what’s going wrong (ID switches, latency, drift, a slow model), and I’ll tell you how I’d approach it.