2025 — PRESENT
Tempus AI
Working across imaging, genomic, and clinical data with multimodal foundation models and agentic systems for oncology.
MACHINE LEARNING & SCIENTIFIC COMPUTING
Devansh Agarwal
Staff Machine Learning Scientist at Tempus AI
I build AI systems that scale. At Tempus, I work on models and agentic systems that bring together medical imaging, genomic data, and clinical information to help improve patient outcomes.
During my physics PhD, I built models to find signals in astronomical images. I carried that work into CT and MRI imaging at Bristol Myers Squibb, and now work across a wider range of biological data.
Work & research
More about my work2021 — 2025
Bristol Myers Squibb
Carrying image-based machine learning into CT and MRI data, with work on self-supervised vision models, tumor assessment, and clinical-trial applications.
2016 — 2020
Physics & radio astronomy
Studying astronomical images and building models to detect faint signals. My PhD research also involved real-time GPU pipelines for telescope data.
Recent publications
View publications- The RRATalog : a Galactic census of rotating radio transients
Monthly Notices of the Royal Astronomical Society
- Twenty-four thousand hours of GREENBURST observations with the GBT
Monthly Notices of the Royal Astronomical Society
- Masked Image Modeling Advances 3D Medical Image Analysis
IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)
- Comprehensive Analysis of a Dense Sample of FRB 121102 Bursts
The Astrophysical Journal
Bibliography via OpenAlex. View Google Scholar
Beyond work
I enjoy cooking and Star Trek. I’ve also led deep learning study and coding sessions with Machine Learning Tokyo.
Teaching materials