Computer Vision Architect

Requirements:

• Strong foundation in computer vision techniques: object detection, image segmentation, classification, pose estimation, OCR, and tracking.
• Experience in fine-tuning pre-trained architectures (ResNet, YOLO, ViT, DETR, SAM, CLIP), choosing the right approach based on data availability and use case.
• Proficiency in deep learning frameworks PyTorch and TensorFlow, along with CV libraries like OpenCV, Detectron2 MMDetection, and Hugging Face Transformers/timm.
• Solid understanding of neural network architectures for vision (CNNs, vision transformers, GANs, and diffusion models) and ability to design or adapt them for specific problems.
• Experience with transfer learning and parameter-efficient fine-tuning (LoRA, adapters) to adapt large/foundation models efficiently.
• Skilled in data pipeline development, collection, annotation (CVAT, Labelbox), augmentation, and synthetic data generation.
• Experience with model optimization for production: quantization, pruning, distillation, and deployment via TensorRT, ONNX, OpenVINO, or on edge devices.
• Working knowledge of MLOps practices, experiment tracking (MLflow, W& B), CI/CD for ML, model versioning, and production monitoring.
• Familiarity with cloud platforms (AWS, GCP, Azure) and containerization (Docker, Kubernetes) for scalable deployment.
• Sensor fusion/multi-modal data experience (LiDAR, depth cameras) is a plus, depending on the domain.

Leadership and Soft Skills:

• Proven experience leading a team of CV/ML engineers, mentoring, code reviews, and technical growth.
• Ability to evaluate build-vs-fine-tune-vs-buy trade-offs and set realistic project roadmaps.
• Track record of shipping models from prototype to production at scale.
• Strong cross-functional collaboration with product, data, and engineering teams.
• Stays current with CV research and emerging pre-trained models, applying them pragmatically.
• Clear communicator of technical trade-offs to both technical and non-technical stakeholders.