AI/ML Scientist, Planetary Science

Relativity Space•Long Beach, CA

1d•Hybrid

About The Position

Relativity Space is seeking an AI/ML Scientist to develop and deploy machine learning systems for their 2028 Mars orbital mission. This role involves working at the intersection of frontier AI methods and planetary science, creating new approaches for analyzing disparate datasets, heterogeneous instrument modalities, and dynamic planetary systems. The scientist will build AI models to run on the spacecraft in Mars orbit, focusing on enhancing Mars atmospheric modeling and weather forecasting, and developing methods for multi-modal data fusion to reconstruct 3D representations from various datasets. The position also includes building systems for real-time observation analysis, scientific event detection, and developing an AI decision-making layer for autonomous spacecraft re-tasking. This is a high-ownership, applied research role requiring creativity, collaboration, and end-to-end system development. The role is jointly advised by Relativity's Interplanetary Sciences Program and Polymathic AI, with collaboration involving travel to New York. The selected candidate will join a vibrant, interdisciplinary team based in Long Beach, CA and New York City.

Requirements

PhD in machine learning, computer science, physics, or a related technical field
Demonstrated experience with transfer learning, domain adaptation or model fine-tuning, particularly in low-data or out-of-distribution settings
Experience with applying machine learning in physical datasets
Working knowledge of multi-modal data fusion
Ability to own problems end-to-end: from dataset understanding through model development, evaluation, and deployment
Excited to collaborate with a diverse group of scientists and engineers, and further planetary science

Responsibilities

Develop and deploy machine learning systems to unlock new science from the 2028 Mars orbital mission.
Build AI models to run on the spacecraft in Mars orbit.
Develop and apply Machine Learning techniques to combine Earth-derived atmospheric datasets and known Martian atmospheric physics to create a weather forecasting model.
Optimize the weather forecasting model to run on the spacecraft at Mars with real-time collected data as input.
Develop and build methods that reconstruct coherent 3D representations by integrating complementary datasets (2D surface images, 3D surface models, geologic mapping, radar depth soundings).
Build systems that monitor observations, analyze them in real-time on the spacecraft, and detect scientifically significant events.
Develop the AI decision-making layer that closes the loop, autonomously re-tasking the spacecraft to acquire follow-up observations from onboard inference on flight hardware.
Drive problem framing, build and evaluate systems end-to-end, and communicate results clearly to scientists and engineers.
Combine core principles of machine learning with deep learning tools to amplify science discovery of the Mars mission.