The contributor platform behind frontier AI.

Compete on hard problems across STEM, earn bounties, and produce the data that drives LLMs forward.

app.shipd.ai

Pre-Submission

Determinism

Score Validation

Overfitting Checkplateau at iter 340

Code Quality

Memory Usage

Ranking

EeigenSolver0.0312

Pprob_master0.0318

Mmathwhiz420.0327

#4you0.0341

Reviews

EeigenSolverREV

min_samples_leaf=5 is too aggressive for this dataset size. Try 10–15 to avoid overfitting on the shallow splits.

Pprob_masterOK

Clean approach with early stopping. The predict helper is well-structured.

Bounty $210

solution.py

test_solution.py

Explorer

challenges

gradient-boost

solution.py

test_solution.py

README.md

config.yaml

workspace/challenges/gradient-boost/solution.py

1import numpy as np

2from sklearn.metrics import mean_squared_error

4def train_gradient_boosted_regressor(

5 X_train, y_train, X_val, y_val,

6 n_estimators=500, lr=0.01, max_depth=4

7):

8 """Gradient boosted regressor with early stopping."""

10 residuals = y_train.copy()

11 trees = []

12 val_scores = []

13 best_val, patience_ctr = np.inf, 0

15 for i in range(n_estimators):

16 tree = DecisionTreeRegressor(

17 max_depth=max_depth,

18 min_samples_leaf=5

19 )

20 tree.fit(X_train, residuals)

21 pred = tree.predict(X_train)

22 residuals -= lr * pred

23 trees.append((lr, tree))

25 val_pred = predict(trees, X_val)

26 val_mse = mean_squared_error(y_val, val_pred)

27 val_scores.append(val_mse)

29 if val_mse < best_val - 1e-4:

30 best_val = val_mse

31 patience_ctr = 0

32 else:

33 patience_ctr += 1

34 if patience_ctr >= 20:

35 break

37 return trees, val_scores

Problems1

OutputTerminal

L13Determinism

L26Score Validation

L29Overfitting Check— plateau at iter 340

L18eigenSolver: min_samples_leaf=5 is too aggressive for this dataset size. Try 10–15 to avoid overfitting on the shallow splits.

L25prob_master: Clean approach with early stopping. The predict helper is well-structured.

community

Where Kaggle grandmasters, OSS contributors, and tinkerers compete

LangChain

React

Airbyte

Next.js

PyTorch

Hugging Face

Top OSS contributors

From the world's most impactful open-source projects.

Kaggle GrandmasterKaggle

Legendary GrandmasterCodeforces

LeetCode GuardianLeetCode

ICPC World FinalistICPC

Top-ranked across platforms

Competitive programmers and ML engineers ranked among the best on every major platform.

Live

2,847

registered contributors

A global community

Engineers from 40+ countries, architecting and solving the challenges that frontier models can't.

Our pool is deep. For each project, we invite top contributors to join — there's something for every skill level. Keep cooking and you'll unlock cooler and cooler projects.

quests

You’re not here to annotate data.
Work on what excites you.

Each unique project on Shipd is called a ‘quest’, and focuses on a specific category in STEM, including SWE, ML, DS, and more. You pick what you want to work on.

Mars

Software Engineering

Tackle GitHub issues, build features, and fix bugs across the repos you care about.

Eris

Machine Learning

Build and solve ML competitions that stump agents and humans alike.

Monte

Data Science

Create and solve numerical computation problems that push AI reasoning to its limits.

Explore

All Disciplines

New quests launch every month across dozens of STEM categories. Create an account to get notified when projects match your skills.

Explore Issues

6 available

Fix memory leak in WebSocket handler

acme/realtime-server

$120

bugperformancehard

Add dark mode toggle to settings page

openui/components

$80

featureuieasy

Implement rate limiting middleware

fastapi/contrib

$150

featuresecuritymedium

Optimize database query for user search

prisma/examples

$100

performancedatabasemedium

Fix SSR hydration mismatch

remix-run/remix

$175

bugssrhard

Add TypeScript types for plugin API

vitejs/vite

$90

typesdxeasy

Browse Challenges

6 active

Code Generation Rubric

Code Quality

$180

14 solvers

Math Reasoning Benchmark

Reasoning

$200

8 solvers

Instruction Following Eval

Alignment

$150

22 solvers

Multi-Turn Dialogue Scorer

Conversation

$175

6 solvers

Vision-Language Grounding

Multimodal

$220

3 solvers

Long-Context Retrieval Eval

Retrieval

$190

11 solvers

Open Problems

6 available

Eigenvalue Decomposition Edge Cases

Linear Algebra

$150

hard3 subs

Stochastic Gradient Descent Convergence

Optimization

$200

expert1 subs

Monte Carlo Integration Variance Reduction

Statistics

$100

medium7 subs

Sparse Matrix Factorization

Linear Algebra

$175

hard2 subs

Fourier Transform Numerical Stability

Signal Processing

$120

medium5 subs

Bayesian Posterior Sampling

Statistics

$200

expert0 subs

The contributor platform behind frontier AI.

Where Kaggle grandmasters, OSS contributors, and tinkerers compete

Top OSS contributors

Top-ranked across platforms

A global community

You’re not here to annotate data.Work on what keeps you up at night.excites you.

Mars

Eris

Monte

Explore

Explore Issues

Browse Challenges

Open Problems

You’re not here to annotate data.
Work on what excites you.