Senior AI Scientist (SF)

At Atomionics, we are building a map of all the resources that will unlock humanity’s abundant future. We need 500% more metal in the next 10 years. These are metals like Copper, Co, Nickel, lithium, which form the bedrock of the infrastructure that needs to be built. But drilling for these metals is wrong 8/9 times. We are building a large planet model to solve this, powered by our proprietary quantum gravimetry technology that can collect 10000x better gravity data. Gravity encompasses all the information of the earth’s crust including resources like metals, groundwater, hydrogen and geothermal.

This position entails spearheading the development of methodologies for modeling Earth's crust using gravity and other geophysics data. A key challenge will be to devise unique and effective approaches, leveraging frontier models to accurately represent complex geological structures.

A significant aspect of this position involves a close collaboration with the CEO. Together, you will determine the strategic direction for product development, identify key milestones, and ensure the research aligns with business objectives. This involves pinpointing the most viable product opportunity stemming from the research. The goal is to swiftly develop and showcase an initial product that demonstrates the potential and real-world applicability of the research findings.

1. Requirements

• PhD (or PhD-equivalent research record) in Computer Science, Electrical Engineering, or a closely related field completed within the last ~3 years.

• 3 + years leading novel research as a post-doc or at a deep-tech startup.

• First-author publications in premier venues such as NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV, or TPAMI-ideally on structured prediction, probabilistic modelling, or large-scale visual recognition.

• Demonstrated ability to translate novel algorithms into working code (e.g., open-sourced research prototype or competition-winning implementation).

• Fluency in modern ML tooling -Python, PyTorch or JAX, version-controlled experiments, GPU/TPU training workflows.

• Strong grounding in statistical learning theory, graphical models, and Deep Learning based models;

• Excellent written and oral communication-able to explain a structured-prediction model to engineers and geologists alike.

2. Core Skills: what you’ll use every week

DomainWhat “great” looks likeStructured & Probabilistic MLFactor-graph / CRF / structured SVM intuition; comfortable extending to 3-D geospatial grids and physics constraints.Hybrid AI × Physics thinkingKnows when to enforce conservation laws in the loss and when to let the network learn.Experimental disciplineDesigns ablation studies, uncertainty calibration, and significance testing.Hands-on Hardware Integration & RoboticsRapid-prototypes sensor interfaces, syncs ML pipelines with on-device firmware, and loves a lab full of oscilloscopes.Collaboration & mentorshipLeads reading groups, reviews PRs for research quality, and coaches junior scientists.

3. Good-to-Have

• Prior work on remote-sensing, climate, or geoscience data (e.g., hyperspectral cubes, DEMs, gravity or magnetics).

• Hands-on with graph neural nets, diffusion models, or PINNs for physical systems.

• Experience with robotics teams, hackathons, or organizing niche maker events

• Patents or industry internships showing an ability to ship research to production

Why this role?

You’ll own the core algorithms that power our planetary foundation model, working shoulder-to-shoulder with quantum physicists and field geologists to turn noisy gravimeter signals into actionable maps-and publish along the way.