Evalda
AINS Edition

Open-Source ML Competition Judging Platform

Securely evaluate untrusted machine learning models in isolated Docker containers, score them on accuracy, size, and latency, and rank teams on a live anonymous leaderboard.

2Active Competitions
3Scoring Metrics
DockerSandboxed Execution

How it works

From your local environment to the leaderboard in three steps.

01

Build

Train your model on the provided dataset using any framework. Implement preprocess, load_model, and predict, the rest is up to you.

02

Submit

Zip your code with a requirements.txt and upload. The automated judge extracts dependencies, installs them, and runs your model in a secure sandbox.

03

Compete

Get scored instantly on accuracy, size, and latency. Watch your rank update on the live leaderboard, then iterate and resubmit to climb higher.

Built for machine learning competitions

Everything you need to focus on what matters, your model.

Sandboxed Execution

Every submission runs in an isolated Docker container with no network access. Your code, your dependencies, zero interference between teams.

Multi-Factor Scoring

Models are scored on Macro F1 accuracy, model size, and inference latency, combined into a single weighted score that rewards efficient solutions.

Real-Time Verdicts

Watch the evaluation stream live via WebSocket as your submission progresses through the judging pipeline, from extraction to final score.

Anonymous Leaderboard

Live rankings that update after every accepted submission. Team names are hidden from other participants until the competition ends.

Any ML Framework

Bring your own Python dependencies. scikit-learn, PyTorch, TensorFlow, XGBoost, whatever fits your approach, install it at submission time.

Invite-Only Access

Secure competitions with invite codes and email whitelisting. Every participant is verified before they can submit or view the leaderboard.

Current competitions

Choose your challenge and start competing.

Ready to compete?

Enter the competition you were invited to. Sign up with your invite code, or log in to continue where you left off.