In-browser judge
Write Python in a real editor and get instant, test-case-level feedback on every run and submit.
LeetCode × Kaggle, for ML
Practice machine learning into mastery.
Katalume is the training ground for machine learning — solve real ML problems in an in-browser judge, compete in contests, and climb to mastery. LeetCode rigor meets Kaggle depth.
healthcare-feature-mean · main.py
def fit_and_predict(X, y, T, k):
preds = []
for t in T:
d = sorted(
(dist(t, x), y[i])
for i, x in enumerate(X))
top = [lbl for _, lbl in d[:k]]
preds.append(vote(top))
return preds Accepted · 8/8 practice tests
Runtime 114msMemory 28MBEasy
Python runs in an isolated browser worker against deterministic practice tests, with instant case-level feedback.
ML-first
Problems built for data science, not generic DSA
Instant
Test-case-level judging on every submit
60 free
A complete launch set before any membership upgrade
Everything you need to get ML-interview ready
One focused platform — solve, compete, learn, and track. No fluff.
Contests & leaderboards
Timed ML competitions with live standings — climb the global ranks as you solve.
Guided learn tracks
Curated paths that map directly to hands-on problems, from foundations to model selection.
Progress that's real
Streaks, acceptance trends, and per-topic coverage computed from your own submissions.
Ready to sharpen your ML edge?
Create an account and solve your first problem in under a minute.
Get started free