UniPat Fellows: Investing $15 Million to Reshape How Science Is Done
UniPat Fellows brings scientific research and model development into the same feedback loop. Fellows work alongside our researchers and engineers to tackle important problems that AI still can’t solve—but should.
- Investment
- $15Mover the next year
- API credits
- $75Kfrontier-model access
- Formats
- 2 waysfull-time or project-based
- Work mode
- Onsite / Remoteflexible options
Why Fellows
Bridging Science and AI, Starting with Questions That Matter
Across disciplines, researchers see opportunities for AI to advance their work. Yet domain researchers and AI teams often work separately, with limited opportunities for sustained collaboration.
Domain researchers
Questions
that matter.
Know how to frame research problems and judge the results.
May needDedicated AI research and engineering support.
AI teams
Models
and tools.
Bring expertise in AI research and engineering.
May needDomain knowledge to define tasks and validate results.
Working together on shared research problems.
We’re launching UniPat Fellows at the UniPat AI Exploration Lab to bring domain researchers and AI teams together around shared research problems.
Fellows bring hard questions and deep domain expertise. Our researchers and engineers work alongside them to turn those questions into tasks for evaluating and training models. As models improve, they can help Fellows advance the research that inspired those tasks.
UniPat AI will invest $15 million over the next year in this effort. Inspired by Bell Labs’ approach to collaborative research, we provide models, compute, and experimental support so researchers can pursue difficult questions together.
The process
How Fellows Work
Each Fellow starts with a problem from their own research. Together, we identify tasks that still depend on expert researchers and explore what it would take for AI to perform them reliably.
- Benchmarks establish a baseline for model performance and reveal where models fall short.
- Interactive environments give models access to specialized tools and simulations, with real-world feedback guiding their next steps.
- Training. The resulting tasks and feedback provide the basis for model training.
- Evaluation. Subsequent evaluations assess whether model performance on these research tasks has improved.
Our scope spans physics, chemistry, biology, medicine, materials, chip design, and beyond. What matters is whether the problem is important—and whether progress can be tested against observable results.
- Physics
- Chemistry
- Biology
- Medicine
- Materials
- Chip design
Current work
What Fellows Are Exploring
Optical chip design
Peking University · Physics
A Fellow from Peking University with a background in physics is exploring automated optical chip design. Models use electromagnetic simulation tools to work toward design objectives, analyze optical fields and device responses, and refine their designs based on simulation feedback.
Antibody design & engineering
Peking University · Pharmacy
A Fellow working in pharmaceutical research at Peking University is investigating automated antibody design and engineering. Models draw on sequence, structural, and experimental data to propose mutations, then use wet-lab feedback to guide the next round of experiments. The research examines whether AI can participate meaningfully in a continuous cycle of design, experimentation, feedback, and redesign.
Chip layout design
UCLA · Chip design
A Fellow from UCLA is exploring automated chip layout design. Models operate professional design tools to generate and modify layouts under complex constraints, using verification and simulation feedback to improve their work.
Probabilistic forecasting
Prediction markets
Other researchers are exploring probabilistic forecasting, including questions drawn from prediction markets. Forecasts are made before outcomes are known and evaluated as events unfold. Across many resolved events, researchers can assess whether models identify useful signals in noisy information and produce reliable, well-calibrated forecasts.
Who We’re Looking For
- Deep expertise, exceptional talent, or distinctive research taste in your field
- Belief in AI’s potential to substantially advance your discipline
- AI as an integral part of your research or work
What We Provide
- Close collaboration with our research and engineering teams.
- Frontier-model API credits valued at $75,000.
- Compute, specialized data, access to domain experts, and research tools.
- Competitive compensation.
Formats
Two Ways to Join
Long-term
Full-time Fellow
Join UniPat AI to pursue long-term research and real-world applications.
3 months or longer
Project-based Fellow
Work with us on a project for three months or longer while continuing your academic or professional commitments.
The question that starts every Fellowship
“What is a problem in your field that AI still can’t solve, but should?”Email fellows@unipat.ai