Machine Learning Engineer

PUSHIT is a sports replay technology company that uses advanced computer vision to capture and highlight moments from weekly games. As a Machine Learning Engineer, you'll design, train, and deploy CV models that recognize sports events and generate highlight videos running on both cloud infrastructure and edge devices.

• Build and improve computer vision models for object detection, action recognition, semantic segmentation, tracking, and event classification on multi-camera sports footage
• Optimize and convert models (TFLite/ONNX/RT-Core) for edge deployment to run in real-time on Raspberry Pi 5 and ARM-based phones
• Own the ML pipeline including data collection, labeling guidelines, experiment tracking with Weights & Biases/MLflow, automated training, and CI/CD to Kubernetes
• Collaborate with backend, video-encoding, and mobile teams to integrate inference results into NestJS APIs and Flutter/Angular clients
• Research state-of-the-art techniques, conduct ablation studies, author technical specifications, and mentor student interns

• 3+ years of professional Python experience with deep learning expertise in TensorFlow and/or PyTorch
• Strong background in computer vision including convolutional networks, attention mechanisms, and spatio-temporal models (e.g., I3D, SlowFast)
• Experience deploying models to mobile or edge hardware (TFLite, Core ML, TensorRT, or similar)
• Familiarity with Docker & Kubernetes workflows for scalable training and inference
• Proven ability to ship production code with at least one project or product in deployment (GitHub, app store, or academic publication)
• Comfortable with version control, code reviews, agile workflows, and technical documentation
• Professional-level English proficiency for global team communication
• Passion for sports or curiosity about sports moments and replays

Schedule & Location: Fully remote

Benefits & Perks: Competitive salary, early equity, budget for conferences and GPU hardware, small team with high ownership and minimal bureaucracy, opportunity to ship features used by thousands of weekend athletes daily