Tech Lead Applied Scientist

About the role

This is the most senior individual contributor position within the Safety AI team, focused on transitioning models from research into production for safety-critical systems. You will bridge the gap between computer vision research and real-world deployment on millions of IoT devices globally. Your work will directly influence safety outcomes for drivers and heavy equipment operators across the physical operations sector.

Key facts

What you'll do

• Lead the design and implementation of real-time perception systems for edge devices and cloud environments.
• Architect end-to-end computer vision pipelines covering object detection, tracking, semantic segmentation, and multi-camera fusion.
• Define the technical roadmap for perception, focusing on model optimization for constrained hardware without sacrificing performance.
• Collaborate with hardware and firmware teams to co-design full-stack solutions.
• Process petabyte-scale multimodal data, including video, telematics, sensor, and diagnostic information.
• Provide technical mentorship to senior scientists and engineers.
• Translate complex business and customer requirements into actionable engineering plans.

Requirements

• Master's or PhD in Computer Science, Robotics, Electrical Engineering, Computer Vision, or a related quantitative field.
• Minimum 10 years of experience as a scientist or ML engineer, with a track record of leading end-to-end AI systems into production.
• Deep technical knowledge of computer vision for real-world environments.
• Experience with sensor fusion and multimodal perception using lidar, camera, radar, GPS, or IMU.
• Proficiency with vision-language models and transformer-based architectures.
• Experience building and deploying ML systems optimized for low-latency edge inference.
• Proven ability to manage the lifecycle of a system from research to production, ensuring reliability and scale.

Nice to have

• Experience applying perception to fleet safety, driver assistance, or operations automation.
• Patents or publications in top-tier venues such as NeurIPS, CVPR, ICCV, or ICRA.

Skills & tools

• Computer Vision (Object detection, tracking, segmentation)
• Multimodal perception and sensor fusion
• Transformer-based architectures
• Vision-language models (VLMs/VLAs)
• Edge ML optimization
• Large-scale training pipelines

Practical notes

• All offers are contingent upon the legal right to work in the United States.
• The company utilizes Tofu for application authenticity verification.
• Official communications will only originate from @samsara.com, @us-greenhouse-mail.io, or @mail3.guide.co domains.
• No fees are charged to applicants at any stage of the hiring process.
• Reasonable accommodations for the recruiting process are available upon request via accessibleinterviewing@samsara.com.