SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Berkshire Grey is a leader in AI and robotics, automating complex pick, pack, sort, and unload operations for e-commerce, retail, and logistics. As a Senior ML Engineer on the Scoop team, you will develop computer vision solutions for trailer-unload robotics, tackling real-world challenges including varied package types, shifting walls, confined spaces, and complex unload conditions.
You will prototype and deploy computer vision approaches that enhance robot perception, reasoning, and interaction with dynamic environments. Your work directly improves system performance, reliability, and throughput while unlocking new customer value. Key responsibilities include developing solution prototypes for computer vision problems, rapidly iterating and creating demos for stakeholders, serving as a subject matter expert during transition to product teams, identifying high-impact improvement areas, staying current with robotics advancements, mentoring junior engineers, and communicating technical priorities.
Minimum qualifications: Master's degree in Robotics, ML, Computer Vision, or CS; 4+ years software development in robotic manipulation; strong Python and C++ expertise; experience with PyTorch and data science tools (numpy, pandas, scipy, matplotlib, scikit-learn); demonstrated experience training and adapting ML architectures (CNNs, ViTs, VLMs) for tasks like grasp estimation, object detection/segmentation, depth estimation, or anomaly detection; proven ability to solve real-world computer vision problems; hands-on experience developing and troubleshooting robotic systems; ability to determine and communicate technical priorities; rapid prototyping and iteration skills; independent project execution; adaptability in fast-paced environments; mentoring capability; and strong communication skills for technical and non-technical audiences.
Preferred: MS/PhD in ML, Computer Vision, or CS; expertise in computer vision for robotic manipulation (grasp estimation, long-tailed object detection in clutter); experience with robotic vision sensors, camera-to-robot calibration, RGB and depth data, real and synthetic dataset collection, ML applied to hardware, real-time perception-based control, robot simulators (Isaac Sim), model-based and data-driven approaches, Docker, cloud computing, experiment tracking (Weights & Biases), database systems (MongoDB), and parallel/distributed systems.
