Tech Lead Applied Scientist

About the Role

Samsara's AI has helped prevent accidents that would have harmed an estimated 380,000 people per year. Every model we ship runs in the real world, on real roads, protecting real drivers.

This is the most senior individual contributor role on our Safety AI team. The role is focused on taking models from research to production in safety-critical systems.

You'll sit at the intersection of cutting-edge computer vision research and production systems that run across millions of IoT devices globally. The models you build run on the edge, in real time, on the trucks, vans, and heavy equipment that keep the world moving.

• Scale: Millions of AI dash cams, 100B+ miles/year (~99% of U.S. roads), 25T+ data points
• Problems: Driver behavior understanding, 3D scene reasoning, and long-tail event detection in real-world environments
• Stack: Modern vision models (e.g. transformers), large-scale training pipelines, and optimization for edge inference on constrained devices

Responsibilities

• Architect end-to-end computer vision pipelines for real-world safety detection — object detection, tracking, semantic segmentation, multi-camera fusion, and beyond
• Drive the technical roadmap for edge and cloud perception, including model optimization and deployment on constrained hardware without sacrificing accuracy
• Partner closely with hardware and firmware teams to co-design the full stack
• Work with petabyte-scale multimodal data (video, sensor, telematics, diagnostics) to train and iterate on production models
• Stay at the frontier of CV and perception research and translate findings into shipped product
• Mentor and technically guide senior scientists and engineers across the team
• Bring clarity to ambiguous problems, translating customer and business needs into precise engineering challenges
• Champion and embed Samsara's cultural principles as the company scales globally

Minimum Requirements

• Master’s or PhD in Computer Science, Electrical Engineering, Robotics, Computer Vision, or a related quantitative field
• 10+ years as a scientist or ML engineer, with experience leading end-to-end AI systems in production
• Deep expertise in computer vision (e.g., object detection, tracking, segmentation) for real-world environments
• Experience with multimodal perception and sensor fusion (e.g., camera, lidar, radar, GPS/IMU)
• Experience with transformer-based architectures and vision-language models (VLMs/VLAs)
• Experience building and deploying real-time or edge ML systems optimized for low-latency inference
• Demonstrated ability to drive systems from research through production deployment, including performance optimization and reliability at scale

Preferred Qualifications

• Experience applying perception to real-world systems such as driver assistance, fleet safety, or operations automation
• Publications or patents at top-tier venues (CVPR, ICCV, ICRA, NeurIPS)