Afaq Saeed

Built a modular Python evaluation framework integrating generated multi-camera video, computer-vision metrics, vision-language models, and automated reporting for autonomous-driving development. • Designed repeatable tests for temporal consistency, cross-camera alignment, semantic correctness, and failure modes, turning research questions into structured engineering checks. • Connected multiple evaluation components into a maintainable workflow using PyTorch, OpenCV, CLIP/DINO-based features, and Azure ML.

Researched and implemented reference-marker detection in fisheye camera images and LiDAR point clouds for a production mobile-mapping workflow. • Built software for image processing, point-cloud filtering, geometric fitting, clustering, and image-to-3D association using Python, OpenCV, PCL, Eigen, and VTK. • Collected and processed real scanner data, created structured test mappings, and evaluated detection and 3D localization accuracy under practical operating conditions.

Integrated vision- and audio-based room-geometry components into a multimodal prototype using NeRF, photogrammetry, CRNN models, Blender, Python, and HPC resources. • Aligned outputs from different sensing modalities, investigated difficult environments such as reflective surfaces and incomplete observations, and documented the resulting system trade-offs. • Worked independently across research software, data collection, visualization, and experimental validation

Led an 8-member engineering team delivering two computer-vision products for automated road inspection and infrastructure asset monitoring. • Owned system architecture across camera acquisition, detection, segmentation, tracking, stereo/SfM reconstruction, GPS-based localization, review tools, and cloud workflows. • Translated product and field requirements into work packages, supported engineers with technical troubleshooting, and kept delivery focused on reliable customer-facing outcomes

• Developed end-to-end Python, OpenCV, and PyTorch software for detecting, measuring, tracking, and geolocating road defects and infrastructure assets from vehicle-mounted cameras. • Integrated visual algorithms with stereo cameras, GPS, 3D reconstruction, data-management tools, and fieldcollected video across approximately 5,000 km of road inspection. • Implemented a Raspberry Pi-based stereo reconstruction prototype and optimized processing choices for constrained hardware and deployment conditions.

• Prototyped computer-vision and 3D reconstruction methods, prepared datasets, and converted successful experiments into reusable components for product development. • Supported testing on changing illumination, camera motion, incomplete detections, and noisy field data rather than relying only on laboratory examples.

Master's in AI with Minor in Robotics

Bachelors in Mechatronics Engineering