whoami

Abhejit Rajagopal

I'm a scientist developing tools to understand, verify, and control AI models, working across text, math, physics, and biology.

How do we know: when is an AI prediction worth trusting? My aim is to connect what a model computes to what our evidence supports.

I was previously a Senior ML Scientist at Altos Labs and a Senior Analyst at Toyon Research. I completed my postdoc at UCSF, working with Peder Larson and Thomas Hope in the Advanced Imaging Technologies Group. I received my PhD from UCSB, working with Shivkumar Chandrasekaran in the Scientific Computing Group. I received my BS from UCLA, working with Fabien Scalzo and with Diana Huffaker in the Integrated NanoMaterials Lab.

news

News

  • Our paper on predicting continuous Alzheimer's outcomes published in Nature Aging!

  • Invited talk at the AI Institute in Dynamic Systems, University of Washington

  • Panel discussion "The New Lab Partner: AI and Future Scientific Discovery" at SXSW 2026 in Austin!

  • scGeneScope published in NeurIPS 2025!

More news (5)
  • Invited talk at Finding Your Inner Model, UC Irvine

  • Invited talk at Dept. of EECS, Washington State Univ.

  • Invited talk at Dept. of Bioengineering, UCLA

  • Invited talk at Dept. of Mechanical Engineering, University of Washington

  • Joined the Allen Institute as Senior AI Research Scientist

Short CV [selected]

Download PDF ↓

Abhejit Rajagopal, Ph.D.

Seattle, WA · Email · Google Scholar · abhe.net

I develop mathematical and computational methods to understand, verify, and control AI models, with applications in imaging, biology, and clinical decision-making.

Current appointments

Education and training

  • 2019–2023 · NIH F32 Postdoctoral Fellow, Radiology & Biomedical Imaging, UC San Francisco.
  • 2014–2019 · Ph.D., Electrical & Computer Engineering, UC Santa Barbara.
  • 2014–2016 · M.S., Electrical & Computer Engineering, UC Santa Barbara.
  • 2010–2014 · B.S., Electrical Engineering, UC Los Angeles.

Selected publications

For a full CV, please get in touch.

wox

Research

Mathematical guarantees. Physical constraints. Biological evidence.

Imaging & inverse problems

How much can we infer from limited measurements?

I combine learned priors with physical constraints to reconstruct images from sparse or noisy data. Applications span PET/MRI, CT, and radar imaging.

relevant papers (3)

Biomedical & clinical AI

Can a predictive model help us choose the next experiment?

I build models linking molecular data, cells, tissue, and medical images to study disease and treatment response. I am interested in how evidence from these models can guide further experiments.

relevant papers (4)