2027 Internship State Estimation, Learned Mapping & Semantic SLAM

Join the Team Bringing Advanced Autonomy to the Built World

At Bedrock, we're moving AI out of the lab and into the real world. Our team includes veterans who helped launch Waymo, scaled Segment to a $3.2B acquisition, and grew Uber Freight to $5B in revenue. Today, we're deploying autonomous systems on heavy construction equipment across the country, improving safety on job sites and accelerating schedules on critical infrastructure projects.

We're not here debating the future of AI. We're deploying it in the real world. In just two years, we've raised $350M and achieved the first fully autonomous excavator deployments in construction.

This is where algorithms meet steel-toed boots. You'll work alongside construction veterans and world-class engineers to solve physical-world problems that simulations can't touch. If you're ready to do meaningful work on hard problems, we'd love to have you join us.

About the Role & Team

Construction sites change with every bucket of dirt. Our State Estimation team builds the maps and localization systems that help autonomous excavators understand where they are and how the terrain is changing.

As an intern on this team, you'll explore how modern learning-based methods can improve our geometry-first mapping stack. That could mean localizing reliably in terrain that looks the same in every direction, building maps that hold up through dust and occlusion, or labeling the map semantically so the machine can tell material to dig from haul roads, spoil piles, and berms. You'll test your ideas on real fleet data, measure them against strong classical baselines, and deliver a prototype the team can build on.

What You'll Do

• Prototype learned SLAM and mapping methods, such as place recognition, odometry, depth completion, and neural occupancy or surface representations
• Fuse lidar and camera segmentation into consistent 3D semantic maps, potentially using vision foundation models or open-vocabulary segmentation
• Develop methods that handle changing terrain, moving material, sparse returns, dust, occlusion, and perceptual aliasing
• Train models on fleet lidar and camera data, and build evaluation pipelines to compare mapping and localization performance against existing methods and ground truth
• Work with perception and planning teams to identify the map properties that matter most for downstream decisions
• Deliver a documented prototype, experimental results, and recommendations for future work

What We're Looking for

Required

• Pursuing a BS, MS, or PhD in computer science, robotics, electrical engineering, or a related field, or equivalent research or industry experience
• Strong Python skills and hands-on model training experience with PyTorch or a similar framework
• Solid understanding of 3D geometry, coordinate frames, and transforms
• Familiarity with SLAM and mapping fundamentals, point clouds, or depth data
• Comfort with messy sensor data and designing experiments that distinguish real improvements from noise

Preferred

• Research or project experience in learned SLAM, semantic mapping, or 3D scene understanding
• Experience with neural scene representations, such as NeRFs, 3D Gaussian splatting, neural occupancy, or signed distance fields
• Experience applying vision foundation models, such as DINOv2 or SAM, to 3D or robotics problems
• Experience with lidar processing or multi-sensor fusion
• Exposure to autonomous vehicle, off-road, or field robotics data
• Familiarity with Rust or C++, and ROS or similar robotics middleware