Research shouldn't be
gated by compute.
Keeboh is the open-source on-ramp to AI-powered research. Sign up, form a team, and launch a Jupyter notebook on a real GPU — in under 15 minutes. No cluster queue, no quota application, no driver setup.
Built for
PhD & MS students
Skip the cluster ticketing queue. Get to your first epoch in one afternoon.
Hackathon teams
48-hour ag-science sprints with pre-loaded USDA corn/soy data and a shared GPU pool.
University cohorts
Course instructors provision a cohort, set per-team hours, and watch progress live.
Launch in minutes
Automated environment setup. Forget Docker headaches and driver versioning; focus entirely on your model architecture.
Pre-loaded datasets
Native access to the UC Davis corn and soy multi-spectral datasets. Validated, cleaned, mounted at /data on every notebook.
Team workspace
Centralized GPU allocation. Manage compute credits and shared experimental results across your entire research lab.
The dataset
USDA corn & soy, ready to train
Multi-spectral field imagery for corn and soybean yield prediction, prepared with UC Davis. Cleaned, split into train/val/test, and mounted read-only at /data on every notebook so your team doesn't burn hours on data plumbing.
- Multi-spectral imagery with yield and crop-condition labels
- Parquet + GeoTIFF, indexed by plot ID
- Shared cache across team members
From soil to silicon
Three simple steps to high-performance compute.
Join your cohort
Sign up with academic email or Google. Pick your cohort, create or join a team.
Launch a notebook
One click. Your session goes from queued → starting → ready in seconds.
Track allocation
Live GPU-hour meter per team. Leads see usage, members see what's left.
The roadmap
Access is step one.
Today, the ability to do frontier research depends on which institution you belong to and which committee approved your grant. Keeboh is building a different path — one where any researcher can get compute, show traction, and earn more.
Access
Pre-configured notebooks on real GPUs, free for academic cohorts. From signup to first experiment in under 15 minutes.
Collaborate
Cross-institution teams, shared notebooks, and a research feed — so researchers can find each other and build on each other's work, not just their own lab's.
Fund
Traction-based research funding. Promising work earns expanded compute allocations and visibility to funders — seed, Series A, Series B for science. Your results decide, not a committee.
Pricing
Free during MVP for university cohorts
We're working with a small group of universities and hackathons through the open beta. Cohorts receive GPU-hour allocations at no cost. We'll publish academic pricing as we approach general availability — our guiding principle is that individual coursework stays free.
Frequently asked
What GPUs do I get?
Notebooks run on shared NVIDIA GPU instances during the MVP, provisioned through our compute partners. Larger-job support (A100/H100 class) will surface in the launch dialog as it becomes available.
Do my notebook sessions persist?
Yes. Your home directory and any files outside /data persist across sessions. The /data mount is shared, read-only, and shared across your team.
What happens to my work after the hackathon ends?
Your notebooks, files, and team workspace stay accessible read-only for 90 days after the cohort closes. You can export everything as a zip at any time.
Is my data private?
Notebooks are scoped to your team via row-level security on the backend. Cohort organizers see aggregate usage but not your code or files.
Is Keeboh really open source?
Yes — Keeboh's platform code will be released under an open-source license ahead of general availability, with a self-host option for institutions running their own GPU clusters. Repo link coming soon; email hello@keeboh.com to get notified.
Our mission
The next decade of breakthroughs won't come only from labs with elite compute clusters. They'll come from researchers everywhere who can move fast across disciplines — if they can get on the ramp. Keeboh exists to democratize research: open source, open access, and a funding path built on traction instead of gatekeepers.