Repository navigation
Home
Agentic Kaggle Arena is a human-directed, agent-assisted initiative for multiple Kaggle challenges. OpenAI Codex, Google Antigravity, and specialist scientific skills help turn hypotheses into tested code, remote training runs, and durable documentation. The human retains control of objectives and approval gates.
| Competition | Metric | Verified result | Full history |
|---|---|---|---|
| Playground Series S6E2 — Predicting Heart Disease | ROC-AUC | Private 0.95535, displayed winner score matched |
[[Playground-Series-S6E2 |
The S6E2 submission was made after the competition closed, so it validates the technical result without changing the official historical ranking.
Every competition gets its own report. It separates local OOF, Public LB, and Private LB evidence; records accepted and rejected experiments; identifies compute and skill usage; and states the reproduction boundary. General methods stay in this Wiki instead of being duplicated across case studies.
Agentic Kaggle Arena — human-directed, evidence-first competitive machine learning with Codex, Antigravity, and specialist skills.