Robotics Engineer (Perception)

Treeswift Inc•New York, NY

4d•$160,000 - $210,000

About The Position

Treeswift empowers energy companies to modernize their field work to meet the growth and challenges ahead by building physical AI for the field worker. Their platform is powered by cutting-edge hardware, sensors (LiDAR, camera, etc.), AI, and software designed to revolutionize work in the toughest environments. Since their first pilot in June 2024, they have grown rapidly, now working with three of the five largest utilities in the US. Their technology has enabled customers to reduce wildfire risk, regulatory and outage risk from vegetation, avoid delays and cost overruns in new construction, and accelerate recovery from severe storms. Treeswift is assembling a team of mission-driven experts with deep industry experience in robotics and enterprise software development. They are headquartered in midtown Manhattan, with additional offices in San Francisco and Philadelphia. They are looking for deeply curious and highly ambitious people who want to have a real-world impact and build the future of field work.

Requirements

5-10 years of experience in perception for robotics
Comfortable in a fast-pace and high-ownership environment.
Excited to collaborate, but also comfortable developing your own opinions.
Proficient in C++ and Python
Extensive experience developing SLAM solutions, ideally involving ground-based LiDAR
Experience with extrinsic and intrinsic sensor calibration, including automated data-driven calibration

Nice To Haves

Prior experience with ultra-wide angle cameras

Responsibilities

Improve the quality and consistency of the product in an increasing breadth of operating conditions.
Lead SLAM development, including incorporating new sensor evolutions and exploring innovative approaches to improving performance in GPS degraded environments.
Accelerate time to delivery by leveling up sensor fusion and calibration.
Fuse multi-user, multi-session point cloud maps.
Automate alignment of LiDAR and imagery data to produce consistent large-scale scenes.
Eliminate the need for extrinsic measurements on vehicle sensors in the field.
Improve the ability to catch and correct potential data quality issues at their source.