Racha Salhi

Racha Salhi - Artificial Intelligence Engineer - Evercam | Himalayas RS Open to opportunities Racha Salhi @rachasalhi Computer vision & AI engineer building production-ready detection, tracking, and OCR systems. What I'm looking for I’m looking for a role where I can lead computer-vision R&D and ship scalable production APIs—combining model evaluation, dataset/labeling pipelines, and real-world camera monitoring—ideally with room to benchmark and iterate fast. I’m an Artificial Intelligence Engineer focused on computer vision R&D and production delivery. At Evercam, I lead research for construction-site monitoring—working on action recognition, multi-object tracking, and detection experiments (including PPE compliance). I also built a camera field-of-view change detection system using LoFTR feature matching to automatically flag camera movement and view drift, and I’ve researched IQA methods to monitor stream health. I pair strong model development with end-to-end evaluation and deployment. I’ve designed object detection evaluation pipelines to drive model selection and release decisions, and I’ve scaled dataset creation with pseudo-labeling using open-vocabulary detectors (OWL-ViT). Previously at Agot.ai, I spearheaded a donut detection + OCR pipeline and improved performance (10% MAPE on counting; 95% OCR accuracy), while fine-tuning detection, keypoint, and segmentation models and cutting labeling costs by 50% through pseudo-labeling. I’m energized by teams that value rigorous benchmarking, clean ML infrastructure, and shipping reliable AI into real camera workflows. Experience Work history, roles, and key accomplishments EV Current Artificial Intelligence Engineer Current Evercam Oct 2024 - Present (1 year 9 months) Led computer vision R&D for construction-site monitoring, including action recognition, multi-object tracking, and PPE compliance detection. Built LoFTR-based camera field-of-view change detection, researched image quali