Each project's training runs. Open & test a live model, or
View & deploy a trained one to make it live. Finished models survive restarts.
👋No projects yet.
Click + New project to upload a CSV and train your first model.
Create a new project
📂Step 1 of 5 — Start a project.
Name it, add an optional description, and upload your CSV. One column should hold the value
you want to predict (the “target”). Arabic and English text supported.
📂
Drop a CSV here or click to browse
The create date is recorded automatically.
Configure training
🎯Step 2 of 5 — Tell the model what to predict.
Choose the target column and a time budget, then Start training.
Want to check your data for errors first? Use Audit data quality (optional).
Columns detected
Preview (first rows)
Data quality audit
Records that look wrong — confidently mislabeled targets, extreme values,
anomalous combinations, incomplete rows — ranked by severity for human review.
Mark each flag as a real error or OK, then export the queue.
Training in progress…
⏳Step 3 of 5 — The engine is working.
It races several algorithms and hundreds of settings, then builds an ensemble of the best.
Watch the live leaderboard. You can leave this page and come back — it keeps running.
Live leaderboard (cross-validated)
Evaluation results
📊Step 4 of 5 — See how good the model is.
These scores are measured on held-out data the model never saw during training, so they’re honest.
Happy with them? Deploy this model to start making predictions.
Confusion matrix (holdout set)
Per-class metrics
Feature importance (permutation, holdout)
i
Model leaderboard
i
Deployment
🚀Step 5 of 5 — Your model is live.
Pick a tab below — each does one thing: Predict one record, Batch test a CSV file,
edit Details & name, or grab the API snippet.
Fields are pre-filled with typical values from your training data — edit any and predict a single record.
Upload a CSV with the same columns as the training data.
Every row gets a prediction; if the file also contains the target column, you get
evaluation metrics on this new data.
Result preview (first rows)
A name creates a stable endpoint /api/models/<name>/predict.
The description is a free-text note for your team. Edit either anytime.
Details
Call this model from your own systems — copy the snippet, or send a batch of records as JSON.