Md Faruk Alam | Computer Vision Engineer

Md Faruk Alam | Computer Vision Engineer I teach machines how to see Computer Vision•Deep Learning•Edge Computing ~ experience Head of Engineering Anuba Technologies May 2025 - Present - Architected a real-time QSR monitoring pipeline on NVIDIA Jetson, processing live RTSP camera feeds with YOLO-based detection, reducing manual monitoring effort by ~70%. - Built a FastAPI + PostgreSQL real-time dashboard backend ingesting inference metadata from Jetson pipelines, serving live monitoring data to a web interface for restaurant operators. Computer Vision Engineer Accu-Label Inc. Dec 2024 - May 2025 - Deployed a fully edge-based CV system on NVIDIA Jetson, detecting apples on a conveyor belt at 9.6 ft/min, calculating precise firing coordinates for an automated labeling machine. - Optimized YOLOv8s segmentation model via TensorRT for on-device inference, achieving sub-100ms latency per frame, and integrated Orbbec 3D depth camera for real-time pixels-per-inch calibration. Computer Vision Research Engineer DFKI (German Research Center for AI) Jun 2024 - Dec 2024 - Developed a custom algorithm to address tracking drift in segmentation-based pipelines, reducing ID switches by ~30% and improving overall tracking consistency. - Trained custom segmentation models for vehicle wheel segmentation, enabling precise inter-vehicle distance estimation by measuring wheel-to-wheel pixel distances. view all 7 experiences → ~ featured projects view all Roboflow Workflow for Tilletia Counting using YOLOv8 Consulting project for building a Roboflow workflow for Tilletia spore counting. Annotated and trained a custom YOLO segmentation model for accurate instance segmentation, optimizing the model training pipeline and workflow within Roboflow. Roboflow RF-DETR vs Ultralytics YOLO26 Comparison Interactive comparison between YOLO26 and RF-DETR object detection models, deployed as a Hugging Face Space for real-time benchmarking. Counting