A SciOS project
Open Compute for Open Science.
Researchers create a computational project. Volunteers donate idle resources. Science moves forward.
There's over 6,000 ExaFLOPS of idle processing in the world. Lettuce use it for science.
§ 01 — Researchers
Stand up a project in less than an hour.
§ 01 — Researchers
Stand up a project in less than an hour.
Describe your computational project, then choose a few settings. We write the code, host the server, distribute the work, and hand back validated results.
Free compute, bounded by the number of people who find your work interesting. The more you convince volunteers to care, the more compute you get. The constraint is attention, not budget.
You set it up once. We handle the rest. No cluster to manage, no cloud bill, no operations team. Validated results land in your dashboard as volunteers complete the work.
Each project sets how its results are checked. You can send the same piece of work to several volunteers in parallel and accept a result only when their independent runs agree.
§ 02 — Volunteers
Does your computer dream of science?
§ 02 — Volunteers
Does your computer dream of science?
Browse active computations, attach to the ones whose questions you want to help answer, leave the rest alone. Run work only when your machine is idle, only on specific projects, or always.
The contribution is yours. A public profile collects what you've worked on, under a handle you choose — pseudonymous is fine. You can edit it, hide it, or take it down at any time.
Credit decisions are signed. The head signs them with its Ed25519 key, so anyone holding the head's public key can check that a signed decision came from that head and has not been altered since it was signed.
Your compute goes to science. Not laundered through a cloud bill, not training a commercial model. Direct contribution to the work you chose, with the results you produce attributed back to you.
§ 03 — Now running
Research active on SciOS Compute.
§ 03 — Now running
Research active on SciOS Compute.
Great Research Extraction Project — translating the scientific literature into frontier research systems. Frontier systems like modular research objects need to know what every paper used and produced — starting with software. GREP's first campaign extracts software mentions from papers: three independently trained models each read every paper's PDF and report the software, libraries, and tools the authors used, with structured type and purpose labels. Each was trained differently, so each makes different mistakes: a mention only enters the record when at least two of the three find it. Each work unit goes to up to three separate volunteers and is accepted once two of them return matching output.
§ 04 — Who
Built and operated by SciOS, a 501(c)(3) nonprofit.
§ 04 — Who
Built and operated by SciOS, a 501(c)(3) nonprofit.
SciOS exists to keep scientific infrastructure in the hands of the people who use it. We don't sell compute, we facilitate it. The source code that runs the server is open. The protocol that pays out credit is open. If we ever stop running it, a peer can take over without losing a single result.
Questions, concerns, or want to host your own head? Reach out at contact@scios.tech.