Visiting Staff Scientist

Planet•San Francisco, CA

2d•Remote

About The Position

We are seeking a distinguished Visiting Staff Scientist to join our AI Research (AIR) team for a one-year sabbatical residency. In this role, you will play a pivotal part in our mission to create a “Queryable Earth” by leading the development of Planet’s proprietary geospatial foundation models (GFMs). While Planet has historically leveraged external models like Google’s RSFM and RemoteCLIP, we are now focused on building in-house models specifically trained on our unique imagery. You will lead research into creating temporally dense embeddings that go beyond static annual summaries, capturing the dynamic and ephemeral nature of our planet—from rapid flooding to disaster impacts. You will collaborate with a multi-disciplinary team of "Planeteers" across space operations, data pipelines, and analytics to co-develop AI/ML solutions that leverage the high spatial resolution and near-daily revisit of PlanetScope data.

Requirements

Distinguished Academic Background: PhD and current Faculty/Professor status in Geospatial Analytics, Computer Science, Remote Sensing, or a related field.
Deep Domain Expertise: 12+ years of experience in remote sensing and satellite image analysis, with a proven track record in building AI-based models for environmental change (e.g., flood-extent, water dynamics).
Multimodal AI Fluency: Extensive experience with foundation models, contrastive learning (CLIP-like models), and multi-model vision-language models (MMVLMs).
Advanced Geospatial Toolkit: Proficiency in multi-sensor integration (Landsat, Sentinel-2, PlanetScope, Sentinel-1) and high-resolution mapping at varying scales (3m, 10m, 30m).
Technical Proficiency: Expert-level Python skills and experience with the scientific stack (xarray, Dask, NumPy, Rasterio, GeoPandas) and deep learning frameworks.
Scale-Minded Research: Experience building automated pipelines for preprocessing and labeling planetary-scale datasets.
Collaborative Spirit: A history of leading research labs and a desire to work in a fast-paced, industrial R&D environment.

Nice To Haves

Specialized Environmental Research: Extensive experience specifically in flood damage quantification and methane-related water dynamics.
Proven Funding & Publication Record: History of leading NASA-funded or similar high-impact geospatial research projects.
Architectural Knowledge: Direct experience fine-tuning or modifying specific GFM architectures like TerraMind or Prithvi.
Hybrid Experience: A mix of deep academic rigor and the ability to prototype rapid-change monitoring tools for operational readiness.

Responsibilities

Develop Planet’s Proprietary GFM: Lead the research and development of a foundation model specifically trained on Planet imagery, incorporating the time-axis to create high-cadence time-series embeddings.
Benchmark Geospatial Architectures: Systematically evaluate and compare existing GFMs (e.g., TerraMind, Prithvi, Clay) against PlanetScope data to assess performance, computational cost, and transferability.
Capture Dynamic Earth Events: Design embeddings and workflows optimized for detecting short-lived, high-impact events such as floods, rapid surface-water expansion, and fire.
Multi-Sensor Integration: Explore the synergy between PlanetScope, Sentinel-1 SAR, and other commercial SAR data to ensure robust time-series analysis even under cloud cover.
Human-in-the-Loop Innovation: Use embeddings to design active learning workflows that prioritize labeling and reduce the annotation burden for time-sensitive mapping tasks.
Academic & Technical Leadership: Publish findings in top-tier journals and present at conferences (e.g., IGARSS, CVPR), highlighting PlanetScope’s unique value in the foundation model ecosystem.
Mentor & Collaborate: Oversee the technical direction of a dedicated postdoc and collaborate with Planet’s research scientists to transition prototypes into operational products.