Senior Machine Learning Engineer, Computer Vision

About Overwatch: Overwatch Imaging is an imagery intelligence technology company working to bring sensor autonomy to time-critical airborne search, detection, tracking and monitoring missions. Our Automated Sensor Operator (ASO) software, native to our own line of purpose-built Smart Sensors and as an upgrade for airborne video gimbal platforms, brings Superhuman Vision and Workload Reduction to users of real-time imagery intelligence systems. We fundamentally believe that modern edge processing and AI-enabled autonomy can perform an important set of time-critical imagery intelligence missions better, faster and more safely than humans can alone. Founded in 2016 in Hood River, Oregon, we support private sector companies as well as federal, state and local agencies around the world with missions ranging from wildfire mapping and disaster response to law enforcement, border security, maritime domain awareness and tactical intelligence. About this Role: Join our dynamic team to build, maintain, and optimize the end-to-end data and model pipelines that power our core Artificial Intelligence systems. This is a hands-on role perfect for an ambitious engineer who is passionate about taking raw data through the entire lifecycle, from scraping and wrangling to production-ready, trained models. This role is remote-friendly, with a preference for candidates who can spend occasional time onsite in Hood River, OR. What You’ll Do: - MLOps and Training Infrastructure - Design and improve training pipelines to support faster, continuous model iteration - Enable scalable experimentation across datasets, model architectures, and training strategies - Improve reproducibility, experiment tracking, and comparison of training runs - Data Pipelines and Dataset Quality - Build and refine workflows for: - Data indexing, curation, and preprocessing - Label quality validation and dataset management - Data exploration for targeted labeling - Develop evaluation and tests that reflect real-world operating conditions - Improve synthetic data workflows to support targeted model improvement - Model Development and Experimentation - Train, evaluate, and iterate on computer vision models for: - Detection, segmentation, classification, tracking - Transformer and multi-modal architectures where appropriate - Continuously explore and evaluate: - new model families and research directions - Domain-specific optimizations for aerial imagery, small-object detection - Incorporate multi-modal inputs including: - Multiple image bands - Platform and sensor metadata (range, angles, telemetry, etc.) - Edge deployment and Optimization - Optimize models for deployment on embedded GPU platforms like Nvidia Jetson - Balance accuracy, latency, and resource constraints for real-world, real-time workflows What You’ll Bring Core Experience - 5+ years of experience in machine learning, with a focus on computer vision - B.S., M.S., or Ph.D. in Computer Science, Engineering, or a related technical field, or equivalent practical experience - Hands-on experience building and improving computer vision or perception models in production or real-world systems, especially for object detection, object tracking - Strong understanding of model evaluation, including selecting metrics, building test sets, analyzing false positives/false negatives, and measuring performance over time for iterative improvement Technical Skills - Experience with PyTorch or Tensorflow - Strong understanding of modern computer vision architectures: - CNNs, transformers (ViTs), and/or multi-model models - OpenCV, NumPy, or similar - Experience building training and data pipelines for medium to large scale datasets - Strong python skills and solid software engineering fundamentals Bonus - Aerial imagery, geospatial data, or remote sensing - Photogrammetry - Multi-modal learning systems - Familiarity with state estimation and tracking methods (Kalman Filters, DeepSORT, ByteTrack, etc) - Experience with performance lang…