Umer Malik

I’m an AI Engineer specializing in computer vision, deep learning, and image processing for medical imaging data. I design, train, and deploy CNN models for diagnosis decision support, including classification plus detection/localization approaches.

At Xylexa Inc., I trained and deployed PyTorch-based medical imaging models for clinical workflows, including a ResNet50 system for binary Chest X-Ray classification achieving over 90% accuracy. I build preprocessing pipelines using PyDicom and OpenCV and collaborate with radiologists to annotate and validate data.

I also focus on reliable deployment and secure integration. I integrate AI pipelines with an ORTHANC PACS server for secure DICOM transfer (SSL/TLS), serve models via REST APIs using FastAPI and Django Rest Framework, and use Celery with RabbitMQ for asynchronous inference.

On the engineering side, I deploy containerized workloads on Azure and strengthen security and compliance—integrating firewalls and encryption and passing ISO27001 and cybersecurity audits. I’m currently advancing weakly supervised chest X-ray disease detection using saliency map generation and FastAI-driven region detection.