{"schemaVersion":"agentdepot.eval_plan.v1","asOf":"2026-10-28","storesIpAddresses":false,"query":"ml","matchMode":"matched_task","taskType":"ml","label":"Autonomous ML Engineering, Data Science & Training Agents","recommendedBenchmarks":[{"id":"mle-bench","name":"MLE-bench","officialUrl":"https://github.com/openai/mle-bench","graderType":"Held-out test split vs medal thresholds"}],"depotBenchmarkSuite":{"type":"data_analytics","tasksApi":"/api/benchmark/tasks?type=data_analytics","submitApi":"/api/benchmark/runs"},"stackPicks":{"evalHarness":{"jtbdId":"public-agent-benchmarks","pickProviderId":"promptfoo","coPickProviderId":"braintrust","runnersUpIds":["braintrust","e2b"]},"sandboxOrRunner":{"jtbdId":"coding-agent-evals","pickProviderId":"e2b","coPickProviderId":"daytona","runnersUpIds":["daytona","modal"]},"ciGate":{"jtbdId":"agent-regression-leaderboards","pickProviderId":"braintrust","coPickProviderId":"langfuse","runnersUpIds":["langfuse","weights-biases"]},"judgeCalibration":{"jtbdId":"cost-latency-per-task-benchmarking","pickProviderId":"langfuse","coPickProviderId":"portkey","runnersUpIds":["portkey","openrouter"]}},"targetMetrics":["held-out test split score vs baseline","zero train/test leakage","GPU-hours + token USD per valid submission"],"antiContaminationRule":"Keep held-out competition test labels outside the sandbox and audit generated training code for validation leakage.","nextCalls":["/api/decide?jtbd=public-agent-benchmarks","/api/benchmark/tasks?type=data_analytics","/api/playbooks/agent-evals"]}