What is Vilvik?
A short tour of what the platform does and who it is for.
Vilvik is an optimization platform in the cloud. You hand it a problem, it searches for good solutions, and you get back the answer plus the data you need to understand how it got there.
The platform is built to host many optimization algorithms. Today, the genetic algorithm is the supported algorithm and the available problem templates are built around it; more algorithms and problem types are on the roadmap. For the genetic algorithm itself we build on an extended version of the open-source PyGAD library. PyGAD's documentation is the authoritative reference for parameter semantics.
Read the PyGAD documentation
How pricing works at a glance. The website is offered as a monthly subscription. The API is credit-based. You top up and each submission deducts credits. See Plans and tiers for the subscription plans and Credits for how API credits work.
You can use Vilvik in three ways:
- From the website, by filling in a form. This is the easiest way to start. There are short forms for common problems and a longer form when you want full control.
- From your own code, by calling our REST API. This is what most teams do once they are happy with the results from the website.
- By sharing a result with a teammate or the public, so others can read what you ran and continue from where you left off.
Every run is a submission. A submission has parameters (population size, number of generations, the function you want to optimise, and so on). When the run finishes you get a result with the best solution found, the fitness chart, downloads, and links you can share.
How a run flows through the platform¶
You Vilvik Worker Result
│ submit │ │ │
│───────────▶│ │ │
│ │ enqueue + bill │ │
│ │─────────────────▶│ run in a │
│ │ │ secure │
│ │ │ environment │
│ │ │──────────────────▶│
│ │ status │ │
│◀───────────│ (and webhooks) │ │
When you submit, the platform reads your parameters, checks your API credit balance if the source is the API, queues the job, and runs it in a secure environment. You can poll the submission, wait for an email, or receive a webhook when it finishes. Once the run finishes, API submissions are billed against your credits based on the compute they actually used. The result is then yours to read, share, or continue.
Who Vilvik is for¶
- Researchers who want to try optimization without setting up a pipeline.
- Engineering teams that need to optimise something repeatedly and want an API instead of a notebook.
- Students learning how optimization algorithms behave on real problems.