Intern, Edge Compute

Planet•San Francisco, CA

5h•$35 - $60•Hybrid

About The Position

Planet's mission is to image the entire world every day, making global change visible, accessible, and actionable. We've successfully captured daily imagery of the Earth, and now we're taking the next bold step: making our spacecraft smarter and more efficient using AI and machine learning. We are seeking a talented AI/ML Intern with an emphasis on geospatial analytics and remote sensing to join our Edge Compute team. This is a high-impact role for an ambitious undergraduate or graduate student to work at the intersection of orbital mechanics, computer vision, and hardware-constrained systems. You will help build the algorithms of our next-generation satellites, moving beyond simple image capture to autonomous vision systems that can detect events and react to dynamic Earth conditions in real-time. The ideal candidate is passionate about AI/ML, autonomy, and squeezing performance out of neural networks to run them in the harsh, resource-constrained environment of space. This is a full-time, hybrid role which will require you to be in our San Francisco, HQ 3 days per week.

Requirements

Currently pursuing or recently completed a degree in Computer Science, Robotics, Computer Engineering, Aerospace Engineering, Electrical Engineering, or a related field.
A solid understanding of deep learning fundamentals, particularly in Computer Vision (CNNs or Vision Transformers).
Proficiency in Python and hands-on experience with at least one major ML framework (e.g., PyTorch, JAX).
The ability to break down complex problems and a strong desire to learn how to deploy models on resource-constrained hardware.
Excellent communication skills with the ability to document technical workflows and explain model trade-offs (e.g., accuracy vs. speed vs power).
A collaborative mindset and the ability to work effectively within a cross-functional team of engineers.

Nice To Haves

Hands-on experience with NVIDIA Jetson devices.
Previous experience with satellite imagery, GIS tools, or remote sensing data.

Responsibilities

Design and train computer vision models (object detection, segmentation, change detection) specifically optimized for satellite imagery and edge deployment.
Experiment with model compression techniques—including quantization, pruning, and knowledge distillation—to ensure high-performance inference on low-power hardware.
Explore algorithms that allow satellites to autonomously identify high-value targets (e.g., ships, wildfires, or cloud-free regions) to optimize tasking and downlink.
Profile and validate model performance (latency, power consumption, memory footprint) on edge hardware targets like NVIDIA Jetson or similar accelerators.
Assist in the curation and augmentation of "space-ready" datasets, accounting for unique orbital challenges like varying off-nadir angles and atmospheric noise.
Build end-to-end prototypes that demonstrate autonomous "closed-loop" systems where ML outputs directly influence satellite actions.
Partner with flight software engineers to transition research prototypes into robust, production-ready code for orbital environments.
Analyze model performance and limitations in resource-constrained environments to propose iterative architectural improvements.