AI Research Scientist

hum.ai; Oregon; Hybrid; USD 150,000 - 210,000; Full time

Benefits offered by this job

Early-stage equity

Health insurance

Dental and vision

Job summary

hum.ai in San Francisco or remote is seeking an AI Research Scientist to own the central model and drive state-of-the-art Earth observation research. You will work with data pipelines handling petabytes of data and collaborate with engineers to train and deploy models for customers.

The role requires a PhD or equivalent, 2+ years of industry AI experience, and a track record of original ideas or publications.

Qualifications

• PhD or equivalent in CS/engineering or a related field.
• 2+ years of industry AI experience.
• Publication record or projects showing original AI ideas.
• Experience scaling models from small experiments to large datasets.

Responsibilities

• Own the model architecture and improvements.
• Apply state-of-the-art techniques to diverse data.
• Define success metrics and benchmarks.
• Run fast experiments and scale winners to petabytes.
• Collaborate with the agent team on model integration and usage in product.
• Set milestones and roadmap with leadership, spotting risks early.
• Publish research and present at conferences.

Skills

Python

PyTorch

Experimentation

Education

PhD in Computer Science or related field

Equivalent experience

Tools

ML frameworks

Job description

AI Research ScientistLocation: San Francisco or remote · Full-time

About us

Hum's mission is to learn how the Earth changes by building a "world model for the entire world": one that represents the dynamics of Earth processes, forecasts what a place looks like next, and makes that understanding useful. Founded at the University of Waterloo and backed by top investors like HF0, Climactic, Onto Ventures and angels from Firebase and OpenTable, we are a small team working on some of the largest unsolved problems.

The role

As our AI Research Scientist, you'll own the model at the center of our system. Our engineering team is building the pipelines that ingest petabytes of earth observation and remote sensing data, and an agent that uses that data and our models to answer customer questions. You'll work closely with our Chief Scientist to decide what the model learns from that data and how, and with our engineers to train and experiment and put your models to good use for our customers and the world.

What you'll do

• Own the architecture. Evaluate our current approach, understand the thinking behind it, and change it where the evidence says to.
• Figure out what the data knows. Apply state-of-the-art techniques to uncover the semantics of large, unusual datasets and turn that into downstream performance, working with domain experts.
• Define success. We aren't chasing existing GeoAI benchmarks. You'll help decide what we measure.
• Go from small to large. Prove ideas in fast, cheap experiments, then scale the winners to petabyte-scale training with our engineers.
• Shape the integration. Work with the agent team on how the model gets called, what it returns, and how to make it more useful in the product.
• Plan the path. Set milestones, roadmap, and resourcing with leadership, and spot risks early.
• Share what you learn. Write up and publish research on earth observation foundation models, and represent our work at conferences and with the broader community.

Requirements

• PhD degree in computer science, engineering, a related field, or equivalent experience.
• 2+ years of industry experience building and shipping AI systems
• Deep understanding of cutting edge AI research, with a track record of original ideas or meaningful contributions shown through publications and/or projects.
• Experience taking ideas from small, exploratory experiments in compute- and data-limited settings to performant, large-scale training runs.
• Comfortable designing models for unconventional data modalities beyond "just text".

Bonus

• Experience training models on non-language data (images, time series, sensor data), ideally geospatial or earth observation, or a background in earth science.
• Background in math or physics, especially where it informs how you model physical systems and sensor data.
• Strong engineering fundamentals: fluent in Python and PyTorch, and able to write efficient, reliable training and inference code, and comfortable using agentic coding tools.

Why us?

• Work that matters for the planet. Our customers in nature conservation, carbon removal, and government use our models to protect ecosystems and respond to a changing climate. We take positive impact seriously, including safety and guardrails, and you'll help us get it right.
• A team worth joining. Founded at the University of Waterloo by PhDs and engineers who've built and shipped at scale, we're small, technical, and collaborative. You'll work alongside people who care about the problem as much as you do.
• Well backed. $10M+ raised from firms like HF0, Climactic, Inovia Ventures, Onto Ventures and angels from Firebase and OpenTable.
• Compensation and upside. Competitive salary, meaningful early-stage equity, health, dental and vision. As one of the first AI hires, your equity stake reflects that.