This product was not featured by Product Hunt yet. It will not be visible on their landing page and won't be ranked (cannot win product of the day regardless of upvotes).
Kelvane
Run untrusted AI models safely in a WebAssembly sandbox
Most WebAssembly runtimes are general-purpose. Kelvane is built for one job: running a neural policy you don't fully trust. Every call runs in a fresh sandbox with a hard memory cap, a per-invocation CPU budget, and zero ambient authority (no files, network, or I/O). The model stays host-owned — the module asks for an inference but never touches the weights. You can hot-swap modules live. Honest scope: it's a small, tested integration of proven parts (Wasmtime, ONNX), not a new technique.
Hi Product Hunt 👋
I'm Rakib. I kept running into the same awkward situation: to run a neural network someone else trained — a downloaded model, an auto-updated policy, something from a third party — you end up running their code, with all the trust that implies. The model does inference, sure, but the wrapper around it can touch your files, your network, whatever the process can reach.
Kelvane is my attempt at a small, honest answer to that. Each inference runs in a fresh WebAssembly sandbox with a hard memory cap, a per-call CPU budget, and zero ambient authority — no filesystem, no network, no I/O. The model stays owned by the host, so the untrusted module can ask for a prediction but never touches the weights or the accelerator. And you can hot-swap a running module without a restart.
Two honest things, because I'd rather you know them up front:
- It's an integration of proven parts (Wasmtime, ONNX), not a new invention — the value is a small, auditable, well-tested assembly aimed at one job.
- It's early (v0.x, toy-scale), but it's genuinely tested: a documented threat model, 29 adversarial test cases, a patched engine, real published benchmarks (including the unflattering ones).
It's open source (Apache-2.0). I'd really value feedback on the sandbox design and where the "run untrusted models safely" framing might overreach. Happy to answer anything!
honestly love that you kept weights host-owned and only ship the inference module — that one design choice makes it actually usable for people who care about model provenance. hot-swap is the cherry on top.
About Kelvane on Product Hunt
“Run untrusted AI models safely in a WebAssembly sandbox”
Kelvane was submitted on Product Hunt and earned 0 upvotes and 2 comments, placing #114 on the daily leaderboard. Most WebAssembly runtimes are general-purpose. Kelvane is built for one job: running a neural policy you don't fully trust. Every call runs in a fresh sandbox with a hard memory cap, a per-invocation CPU budget, and zero ambient authority (no files, network, or I/O). The model stays host-owned — the module asks for an inference but never touches the weights. You can hot-swap modules live. Honest scope: it's a small, tested integration of proven parts (Wasmtime, ONNX), not a new technique.
Kelvane was featured in Open Source (68.6k followers), Developer Tools (515.9k followers), Artificial Intelligence (473.8k followers) and GitHub (41.3k followers) on Product Hunt. Together, these topics include over 221.9k products, making this a competitive space to launch in.
Who hunted Kelvane?
Kelvane was hunted by Rakib. A “hunter” on Product Hunt is the community member who submits a product to the platform — uploading the images, the link, and tagging the makers behind it. Hunters typically write the first comment explaining why a product is worth attention, and their followers are notified the moment they post. Around 79% of featured launches on Product Hunt are self-hunted by their makers, but a well-known hunter still acts as a signal of quality to the rest of the community. See the full all-time top hunters leaderboard to discover who is shaping the Product Hunt ecosystem.
Want to see how Kelvane stacked up against nearby launches in real time? Check out the live launch dashboard for upvote speed charts, proximity comparisons, and more analytics.
Rakib