AI/Edge Compute Engineer

Planet•San Francisco, CA

2h•Hybrid

About The Position

Planet is working on the next generation small satellite bus that will host a high resolution optical imaging payload (Pelican) and a hyperspectral payload (Carbon Mapper). We are seeking an AI/Edge Compute engineer to help develop onboard compute and analytics effort on the next generation satellites. . You will work with various teams to finalize the system requirements, develop functional specifications, and down-select the processors. You will work independently and with others on the development of onboard processing and machine learning algorithms for various Earth Observation applications.. You will collaborate with electrical engineers, camera engineers, firmware engineers, data pipeline, and analytics engineers to build and iterate on the solution. This is a full-time, hybrid role which will require you to work from our San Francisco HQ 3 days per week.

Requirements

4+ years developing algorithms and solving problems pertaining to image processing, computer vision, object detection, machine learning
Bachelor’s degree in Computer Science, Computer Engineering, Electrical Engineering, or other relevant degree
Experience in software development, especially for computationally constrained environments
Extensive experience programming in Python and/or C++ with computer vision libraries
Understanding of version control (eg git) and CI tools
Proficient in GPU-based computing frameworks (OpenCL/CUDA/TensorFlow), including working with NVIDIA's TensorRT
Understanding of embedded and high efficiency software architectures
Experience designing low latency, highly performant code
Experience in hardware and software co-design

Nice To Haves

Experience with machine learning on earth observation data sets
Experience with deployment environments for embedded devices, CI, and the cloud in machine learning scenarios
Background in Visual Large Language Models, Foundation Models, zero/few shot learning, and open vocabulary detectors
Low-level programming experience in CUDA/OpenCL
Excellent communication and analytical skills
Ability to work well both autonomously and collaboratively
Additional education or certifications in in a relevant field of study

Responsibilities

Develop onboard computer vision and machine learning models that enable real time segmentation, feature detection, object recognition, and geolocation.
Develop pre-processing of raw imagery onboard the satellite to make them useable by the machine learning algorithms
Collaborate with other space systems and software engineering teams to implement deep-learning-based object detection systems on pre-processed imagery for identifying, characterizing, and classifying objects
Explore and develop new methods for improving detection performance and improving the runtime of inference targeting the NVIDIA Jetson platform.
Work with internal customers to develop onboard processing and machine learning pipelines for various use cases and classification tasks
Help to architect and develop the data pipeline and platform supporting onboard processing
Hardware/software co-development of edge computing solutions for Planet’s next generation spacecraft
Own and drive performance metrics (e.g. latency, memory, bandwidth, power)