AI Expert | Computer Vision & Machine Learning Researcher & Developer at Rival Solutions

Automated detection, segmentation, and recognition of pavement defects, lane markings, and traffic signs to support Pavement Condition Index computation using deep learning pipelines. Calculated real-world 3D road coordinates via camera calibration and geometric projections. Estimated camera pitch angle with vanishing point detection. Developed encoder-decoder neural networks for vanishing point prediction. Generated synthetic images for data augmentation and robustness. Built tools for spherical panorama processing and annotation transformations. Fine-tuned diffusion models for synthetic defect generation. Developed efficient annotation tools and a robust OCR pipeline integrating Tesseract and EasyOCR. Customized Detectron2, NVIDIA Semantic Segmentation, and proprietary neural networks for multi-GPU scalability and added multi-task learning capabilities. Applied super-resolution techniques and extended tracking models for multi-object tracking.

AI Expert | Computer Vision & Machine Learning Researcher & Developer at Faraadid Vision

Developed multiple AI-driven computer vision applications including fire and smoke detection, face recognition, crowd estimation, intrusion detection, pedestrian detection, people counting, vehicle license plate recognition, camera tampering detection, abandoned object detection, and fall detection using 3D CNNs. Built an IPFIX-based network anomaly detection model. Designed and built linear and switching power supplies for FM radio transmitters. Led a computer vision team to develop real-time surveillance and monitoring solutions.

AI Expert | Computer Vision & Machine Learning Researcher & Developer at Rival Solutions

Led development of AI-based deep learning pipelines for automated detection, segmentation, and recognition of pavement defects, lane markings, and traffic signs. Utilized camera calibration and geometric projections for 3D road coordinate estimation. Developed novel encoder-decoder neural networks for vanishing point prediction and enhanced training datasets with synthetic traffic signs and occlusions. Applied super-resolution and multi-object tracking techniques to improve recognition accuracy. Customized and fine-tuned state-of-the-art models and annotation tools to optimize performance and scalability in road element analysis, including OCR pipelines for MUTCD sign text extraction.

AI Expert | Computer Vision & Machine Learning Researcher & Developer at Faraadid Vision

Developed a suite of AI-driven computer vision applications including fire and smoke detection, real-time face recognition, crowd estimation, intrusion detection, pedestrian detection, and people counting systems. Implemented OCR-based vehicle license plate recognition and camera tampering detection. Created fall detection models using 3D CNNs for health monitoring and network anomaly detection models using encoder-decoder architectures. Additionally, designed and built power supplies for FM radio transmitters.

AI Expert | Computer Vision & Machine Learning Researcher & Developer at Faraadid Vision

Developed fire and smoke detection, real-time face recognition, crowd estimation, intrusion detection, pedestrian detection, people counting, license plate recognition, camera-tampering detection, abandoned object detection, 3D CNN-based fall detection, and IPFIX-based network anomaly detection; built surveillance and safety systems for both indoor and outdoor environments.