Optimize
anything.
Vilvik runs evolutionary optimization on your decisions, models, and games, in the browser, no setup.
How it works
Three steps from problem to solution
1. Define your problem
Express the goal as a Python fitness function and give Vilvik the search space.
2. Run evolution
Today, our genetic algorithm explores millions of candidates — in your browser or on the server. More algorithms are coming.
3. See the solution
Get the best result, the convergence curve, and runnable code you can hand to a teammate.
Features
Pre-built applications, ready to optimize
Drop in a Vilvik app and you're a parameter sweep away from a working optimizer. Every app is a recipe with knobs you can turn — mutation, crossover, fitness, search space — so you graduate from copy-paste to your own algorithm at your own pace.
Customizable Parameters
Tune mutation, crossover, fitness, and search space without leaving the page.
Unleash Your Creativity
Drop in your own Python and run it in the same sandbox — no boilerplate.
OPTIMIZATION HUB
One place for every experiment
Run, compare, and iterate on every optimization run from a single dashboard. Adjust the fitness function, mutation, or crossover and see immediately how each tweak moves the best solution — with full history so you can roll back to what was working.
MACHINE LEARNING
Tune the models you already use
Plug Vilvik into Keras, PyTorch, scikit-learn, or anything that exposes a Python entry point. Search hyperparameters, architectures, and training schedules with optimization instead of grid sweeps — reach better models with fewer trials.
LIVE VISUALIZATIONS
Watch every generation, live
Streaming evolution charts update each generation, so you catch stalls, divergence, and breakthroughs the moment they happen. Drill into completed runs with interactive plots that compare populations, fitnesses, and decisions side by side.
MULTI-OBJECTIVE
Optimize one goal — or many
When trade-offs matter (cost vs. accuracy, risk vs. return, weight vs. strength), Vilvik returns a whole Pareto front instead of a single answer. Inspect each non-dominated solution and pick what fits the moment.
ISOLATED SANDBOX
Private by design
Every run executes in a sealed sandbox. Your fitness function, your data, your generated solutions — none of it leaks across runs or to other users. Reproducible, scoped, and yours alone.
Pricing
Start free. Scale on Premium.
Every new account gets a 15-day Premium trial — the full toolset with no card on file. After that, the Free tier still keeps the basics: signup takes 30 seconds. Power users upgrade for higher daily limits and full-history runs.
Free plan
For learning and side projects
- 5 submissions / day with up to 120 s wall time and 384 MB memory.
- GA caps — up to 100 generations, 200 population, 200,000 total fitness evaluations.
- All pre-built apps — SOO, MOO, knapsack, clustering, sklearn, Keras, PyTorch, TSP.
- All demos, free — Genetic Artist, Walker, and the live agent games.
- Runs in your browser — private sandbox per run, no install.
Premium
Higher caps, full history
- 200 submissions / day with up to 600 s wall time and 1024 MB memory.
- GA caps — up to 10,000 generations, 2,000 population, 2,000,000 total fitness evaluations.
- Run more at once — up to 5 concurrent runs and 16 parallel workers each.
- Full-history runs — keep every solution from every generation for analysis.
- Email support and early access to new optimizers.
Enterprise
Need something built for your team?
For organisations that need bespoke algorithms, priority compute, or dedicated onboarding.
- Custom algorithm development
- Priority access & compute
- Training & onboarding for your team
- Direct line to the engineers
Get in touch
We typically respond within one business day.
Stay connected
Follow vilvik
Toronto · usually around 9 am – 6 pm ESTSend us a message
Drop us a line — we'd love to hear from you.
Find us on the map
Get directions133 Richmond St W, Ste 207 #60
Toronto, ON M5H 2L3
We use cookies. Essential cookies keep you signed in. With your consent, optional cookies help us improve the site. Privacy policy.