Distributed by design
Open science for
an unseen universe.
DeepLense is a distributed, open-source research group connecting physics, astronomy, and machine learning across institutions and career stages.
Our mission
Make advanced scientific AI open, rigorous, and useful.
We focus on strong gravitational lensing and dark-matter studies, while developing methods that matter across scientific machine learning: physical priors, limited labels, domain shift, generative models, and verifiable agents.
DeepLense is part of ML4SCI, the nonprofit open-source umbrella that brings together researchers and contributors to apply machine learning to frontier problems in science.
Code, datasets, and research artifacts designed for reuse.
Clear experimental boundaries and honest negative results.
Research grows through sustained, hands-on open-source collaboration.
People
A small core, a wide orbit.
Mentors and collaborators span astrophysics, fundamental physics, and machine learning.
Principal investigators
Principal Investigator · Brown University
Stephon Alexander
Theoretical physicist working across cosmology, quantum gravity, machine learning, and the foundational questions that connect them.
Research team
Google Summer of Code 2026
The current contributor cohort.
Seven projects extend the program from real-lens finding to foundation models, quantum representations, neural operators, super-resolution, and agentic simulation.
Official GSoC listing ↗Support
Research needs an ecosystem.
DeepLense combines public research support, institutional collaboration, and sustained open-source mentorship.
National Science Foundation
SLINGSHOT
“Decoding Dark Matter through Gravitational Lensing” supports simulation, anomaly detection, property inference, and open astronomical tooling.
- Award
- 2108645
- Awarded
- $371,954
- Program
- 2021—2026
Open-source mentorship
Google Summer of Code
DeepLense has grown through multi-year, mentored projects that turn public code into publishable scientific work.
Institutional home
The University of Alabama
Support from Physics & Astronomy connects the program to teaching, research, and the broader SLINGSHOT collaboration.
Join the work
Start with the science.
Stay for the community.
The fastest way into DeepLense is through the open repository: reproduce a result, improve the pipeline, or take on a documented research task. For collaborations, questions, or contributor inquiries, email ml4-sci@cern.ch.