{"id":108712,"url":"https://github.com/jxucoder/awesome-agentic-machine-learning","name":"awesome-agentic-machine-learning","description":"A curated list of awesome agentic machine learning projects, frameworks, and resources.","projects_count":111,"last_synced_at":"2026-07-22T19:00:20.079Z","repository":{"id":330404532,"uuid":"1122641229","full_name":"jxucoder/awesome-agentic-machine-learning","owner":"jxucoder","description":"A curated list of awesome agentic machine learning projects, frameworks, and resources.","archived":false,"fork":false,"pushed_at":"2026-05-22T23:11:44.000Z","size":45,"stargazers_count":1,"open_issues_count":4,"forks_count":2,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-07-04T04:03:45.163Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":"","language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/jxucoder.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2025-12-25T07:28:00.000Z","updated_at":"2026-05-22T23:11:47.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/jxucoder/awesome-agentic-machine-learning","commit_stats":null,"previous_names":["jxucoder/awesome-agentic-ml","jxucoder/awesome-agentic-machine-learning"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/jxucoder/awesome-agentic-machine-learning","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jxucoder%2Fawesome-agentic-machine-learning","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jxucoder%2Fawesome-agentic-machine-learning/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jxucoder%2Fawesome-agentic-machine-learning/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jxucoder%2Fawesome-agentic-machine-learning/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/jxucoder","download_url":"https://codeload.github.com/jxucoder/awesome-agentic-machine-learning/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jxucoder%2Fawesome-agentic-machine-learning/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":35773465,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-07-20T02:08:10.276Z","status":"online","status_checked_at":"2026-07-22T02:00:06.236Z","response_time":124,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"created_at":"2026-01-17T18:48:30.727Z","updated_at":"2026-07-22T19:00:20.079Z","primary_language":null,"list_of_lists":false,"displayable":true,"categories":["License","Research Papers","Datasets \u0026 Benchmarks","AutoML Agents","Frameworks \u0026 Platforms","MLE-bench Leaderboard","Related Resources"],"sub_categories":["Foundation Models for ML","Benchmarks \u0026 Evaluation","Multi-Agent Systems","Search \u0026 Planning Methods","Autonomous Data Science Agents","Domain-Specific Agentic ML","Surveys","LLM-Based ML Optimization"],"readme":"# Awesome Agentic ML [![Awesome](https://awesome.re/badge.svg)](https://awesome.re)\n\nAgentic ML refers to autonomous AI systems that can plan, execute, and iterate on machine learning workflows with minimal human intervention—from data preprocessing to model training, evaluation, and deployment.\n\n🤖 *This resource list is maintained with the help of [Claude](https://www.anthropic.com/claude) by Anthropic.*\n\n---\n\n## Scope: What Counts as Agentic ML?\n\nA resource belongs in this list when it demonstrates autonomous, closed-loop progress on machine learning workflows (not just general-purpose agent behavior).\n\n**Inclusion checklist (meet at least 3 of 4):**\n\n- Goal-directed ML planning (selects or revises strategy for an ML objective)\n- Tool-using execution across ML lifecycle stages (data prep, feature engineering, model training, evaluation, experimentation)\n- Iterative feedback loop (uses metrics/errors/results to revise actions)\n- Empirical ML outcome (benchmark, competition result, ablation, or measurable improvement)\n\n**Exclude or deprioritize:**\n\n- Generic web/desktop/coding agents without meaningful ML workflow contribution\n- Social-only announcements without a primary technical source (paper/repo/blog)\n- Stale resources unless there is a substantive new release/update in the recent window\n\n---\n\n## Research Assistant Agent\n\nThis repository runs a weekly **Research Assistant Agent** via GitHub Actions to scout and triage potential additions.\n\n- Workflow: `.github/workflows/weekly-resource-research.yml`\n- Primary curation model: `gpt-5.3-codex` with `xhigh` effort\n- Default Grok scout model: `grok-4-1-fast-reasoning` (override with `GROK_MODEL` repository variable)\n- Signal sources: xAI Grok social scout + arXiv RSS scout + Codex curation pass with live web search validation\n\n**Behavior:**\n\n- If high-confidence additions are found, the agent updates `README.md` and opens a draft PR with a supporting suggestion log.\n- If no high-confidence additions are found, the agent opens an issue log with the weekly scout outputs (instead of forcing changes).\n- The agent applies the inclusion checklist in this README and avoids generic non-ML-workflow agent news.\n\n---\n\n## Contents\n\n- [Research Assistant Agent](#research-assistant-agent)\n- [Frameworks \u0026 Platforms](#frameworks--platforms)\n- [AutoML Agents](#automl-agents)\n- [Research Papers](#research-papers)\n  - [Benchmarks \u0026 Evaluation](#benchmarks--evaluation)\n  - [Autonomous Data Science Agents](#autonomous-data-science-agents)\n  - [Multi-Agent Systems](#multi-agent-systems)\n  - [Search \u0026 Planning Methods](#search--planning-methods)\n  - [Domain-Specific Agentic ML](#domain-specific-agentic-ml)\n  - [LLM-Based ML Optimization](#llm-based-ml-optimization)\n  - [Surveys](#surveys)\n  - [Foundation Models for ML](#foundation-models-for-ml)\n- [Datasets \u0026 Benchmarks](#datasets--benchmarks)\n- [MLE-bench Leaderboard](#mle-bench-leaderboard)\n- [Related Resources](#related-resources)\n- [Contributing](#contributing)\n\n---\n\n## Frameworks \u0026 Platforms\n\n*End-to-end platforms and frameworks for building agentic ML systems.*\n\n| Project | Description | Stars |\n|---------|-------------|-------|\n| [AlphaEvolve](https://deepmind.google/blog/alphaevolve-a-gemini-powered-coding-agent-for-designing-advanced-algorithms/) | Google DeepMind's evolutionary coding agent for scientific and algorithmic discovery using Gemini. | - |\n| [AutoGluon](https://github.com/autogluon/autogluon) | Open-source AutoML toolkit by Amazon with foundational models and LLM agents. | ![GitHub stars](https://img.shields.io/github/stars/autogluon/autogluon?style=flat-square) |\n| [EvoAgentX](https://github.com/EvoAgentX/EvoAgentX) | Open-source framework for building, evaluating, and evolving LLM-based agentic workflows. EMNLP 2025. | ![GitHub stars](https://img.shields.io/github/stars/EvoAgentX/EvoAgentX?style=flat-square) |\n| [EvoMaster](https://github.com/sjtu-sai-agents/EvoMaster) | Foundational auto-research agent framework for self-evolving scientific agents, including ML-Master and MLE-bench Lite workflows. | ![GitHub stars](https://img.shields.io/github/stars/sjtu-sai-agents/EvoMaster?style=flat-square) |\n| [Karpathy](https://github.com/K-Dense-AI/karpathy) | Agentic ML Engineer using Claude Code SDK and Google ADK. By K-Dense. | ![GitHub stars](https://img.shields.io/github/stars/K-Dense-AI/karpathy?style=flat-square) |\n| [K-Dense Web](https://k-dense.ai/) | Autonomous AI Scientist platform with dual-loop multi-agent system for research, coding, and ML. | - |\n| [LangGraph ML Flows](https://blog.langchain.dev/langgraph-ml-workflows/) | LangGraph patterns for agentic ML experimentation loops (hyperparameter tuning, evaluation, and iterative refinement). | - |\n| [OpenEvolve](https://github.com/codelion/openevolve) | Open-source implementation of Google DeepMind's AlphaEvolve for evolutionary code optimization. | ![GitHub stars](https://img.shields.io/github/stars/codelion/openevolve?style=flat-square) |\n\n---\n\n## AutoML Agents\n\n*LLM-powered agents for automated machine learning pipelines.*\n\n| Project | Description | Stars |\n|---------|-------------|-------|\n| [Agent Laboratory](https://github.com/SamuelSchmidgall/AgentLaboratory) | Autonomous research framework with specialized agents for literature review, experimentation, and report writing. | ![GitHub stars](https://img.shields.io/github/stars/SamuelSchmidgall/AgentLaboratory?style=flat-square) |\n| [AIBuildAI](https://github.com/aibuildai/AI-Build-AI) | Hierarchical agent for automatically building AI models from task descriptions and training data. | ![GitHub stars](https://img.shields.io/github/stars/aibuildai/AI-Build-AI?style=flat-square) |\n| [AIDE](https://github.com/WecoAI/aideml) | AI-powered data science agent using tree search for solution exploration. | ![GitHub stars](https://img.shields.io/github/stars/WecoAI/aideml?style=flat-square) |\n| [AIRA-dojo](https://github.com/facebookresearch/aira-dojo) | Meta's AI research agents using search policies (Greedy, MCTS, Evolutionary). | ![GitHub stars](https://img.shields.io/github/stars/facebookresearch/aira-dojo?style=flat-square) |\n| [AiScientist](https://github.com/AweAI-Team/AiScientist) | File-as-Bus long-horizon ML research agent for paper reproduction and Kaggle-style MLE workflows. | ![GitHub stars](https://img.shields.io/github/stars/AweAI-Team/AiScientist?style=flat-square) |\n| [AutoGluon Assistant](https://github.com/autogluon/autogluon-assistant) | Multi-agent system for end-to-end multimodal ML automation. Also known as MLZero. | ![GitHub stars](https://img.shields.io/github/stars/autogluon/autogluon-assistant?style=flat-square) |\n| [AutoMind](https://github.com/zjunlp/AutoMind) | Adaptive agent with expert knowledge base from 455 Kaggle competitions and tree search. By ZJU NLP. | ![GitHub stars](https://img.shields.io/github/stars/zjunlp/AutoMind?style=flat-square) |\n| [AutoML-Agent](https://github.com/DeepAuto-AI/automl-agent) | Multi-Agent LLM Framework for Full-Pipeline AutoML. | ![GitHub stars](https://img.shields.io/github/stars/DeepAuto-AI/automl-agent?style=flat-square) |\n| [Data Interpreter](https://github.com/geekan/MetaGPT) | LLM agent for data science using hierarchical graph modeling and programmable node generation. Part of MetaGPT. ICLR 2025. | ![GitHub stars](https://img.shields.io/github/stars/geekan/MetaGPT?style=flat-square) |\n| [DS-Agent](https://github.com/guosyjlu/DS-Agent) | Automated data science agent using case-based reasoning from Kaggle. ICML 2024. | ![GitHub stars](https://img.shields.io/github/stars/guosyjlu/DS-Agent?style=flat-square) |\n| [FM Agent](https://github.com/baidubce/FM-Agent) | Baidu's foundation model agent for ML engineering tasks. | ![GitHub stars](https://img.shields.io/github/stars/baidubce/FM-Agent?style=flat-square) |\n| [FeatureForge](https://github.com/featureforge-ai/featureforge) | Agentic feature engineering toolkit automating feature generation/extraction with reported performance gains. | ![GitHub stars](https://img.shields.io/github/stars/featureforge-ai/featureforge?style=flat-square) |\n| [FOREAGENT](https://github.com/zjunlp/predict-before-execute) | Predict-then-Verify ML agent using run-free preference prediction to reduce expensive execution loops. | ![GitHub stars](https://img.shields.io/github/stars/zjunlp/predict-before-execute?style=flat-square) |\n| [InternAgent](https://github.com/Alpha-Innovator/InternAgent) | ML engineering agent with DeepSeek-R1 integration. | ![GitHub stars](https://img.shields.io/github/stars/Alpha-Innovator/InternAgent?style=flat-square) |\n| [LADS (LightAutoDS)](https://github.com/sb-ai-lab/LADS) | Multi-AutoML agentic system combining LLM code generation with AutoGluon, LightAutoML, and FEDOT. | ![GitHub stars](https://img.shields.io/github/stars/sb-ai-lab/LADS?style=flat-square) |\n| [MLE-STAR](https://research.google/blog/mle-star-a-state-of-the-art-machine-learning-engineering-agents/) | Google's ML engineering agent using web search and targeted code block refinement. Built with ADK. | - |\n| [ML-Master](https://github.com/sjtu-sai-agents/ML-Master) | AI-for-AI agent integrating exploration and reasoning with adaptive memory. By SJTU SAI. | ![GitHub stars](https://img.shields.io/github/stars/sjtu-sai-agents/ML-Master?style=flat-square) |\n| [OpenHands](https://github.com/All-Hands-AI/OpenHands) | Open-source AI software development agent adaptable to ML tasks. | ![GitHub stars](https://img.shields.io/github/stars/All-Hands-AI/OpenHands?style=flat-square) |\n| [R\u0026D-Agent](https://github.com/microsoft/RD-Agent) | Microsoft's research \u0026 development agent for ML tasks. | ![GitHub stars](https://img.shields.io/github/stars/microsoft/RD-Agent?style=flat-square) |\n| [SELA](https://github.com/geekan/MetaGPT/tree/main/metagpt/ext/sela) | Tree-Search Enhanced LLM Agents for AutoML using MCTS. Part of MetaGPT. | ![GitHub stars](https://img.shields.io/github/stars/geekan/MetaGPT?style=flat-square) |\n\n---\n\n## Research Papers\n\n*Academic papers on agentic ML, autonomous ML systems, and LLM-based ML agents.*\n\n### Benchmarks \u0026 Evaluation\n\n*Papers introducing benchmarks and evaluation methodologies for agentic ML systems.*\n\n- **MLE-bench: Evaluating Machine Learning Agents on Machine Learning Engineering** (2024) - [Paper](https://arxiv.org/abs/2410.07095) | [Code](https://github.com/openai/mle-bench)  \n  Benchmark by OpenAI with 75 Kaggle competitions for evaluating ML engineering agents.\n\n- **MLE-Smith: Scaling MLE Tasks with Automated Multi-Agent Pipeline** (2025) - [Paper](https://arxiv.org/abs/2510.07307)  \n  Automated pipeline transforming raw datasets into competition-style MLE challenges.\n\n- **MLAgentBench: Evaluating Language Agents on Machine Learning Experimentation** (ICML 2024) - [Paper](https://openreview.net/forum?id=1Fs1LvjYQW)  \n  Benchmark for evaluating LLM agents on ML research tasks including model training and debugging.\n\n- **MLR-Bench: Evaluating AI Agents on Open-Ended Machine Learning Research** (2025) - [Paper](https://arxiv.org/abs/2505.19955)  \n  Benchmark with 201 research tasks from NeurIPS, ICLR, and ICML. Includes MLR-Judge for automated evaluation.\n\n- **DataSciBench: An LLM Agent Benchmark for Data Science** (2025) - [Paper](https://arxiv.org/abs/2502.13897) | [Code](https://github.com/THUDM/DataSciBench)\n  Comprehensive benchmark with Task-Function-Code (TFC) framework for rigorous evaluation of LLMs on data science tasks.\n\n- **LMR-BENCH: Can LLM Agents Reproduce NLP Research?** (EMNLP 2025) - [Paper](https://aclanthology.org/2025.emnlp-main.314/)\n  Tasks LLM agents with reproducing functions from NLP research papers. Tests understanding of scientific methods and implementation.\n\n- **DS-1000: A Natural and Reliable Benchmark for Data Science Code Generation** (2022) - [Paper](https://arxiv.org/abs/2211.11501)\n  1,000 data science problems from StackOverflow covering NumPy, Pandas, SciPy, Scikit-learn, Matplotlib, PyTorch, and TensorFlow.\n\n- **MLE-Dojo: Interactive RL Environment for Machine Learning Engineering** (2025) - [Paper](https://arxiv.org/abs/2505.07782)\n  Transforms MLE-bench into a Gym-style RL environment with 200+ Kaggle competitions, enabling agent training via supervised fine-tuning and reinforcement learning.\n\n- **Agent^2 RL-Bench: Can LLM Agents Engineer Agentic RL Post-Training?** (2026) - [Paper](https://arxiv.org/abs/2604.10547) | [Code](https://github.com/microsoft/RD-Agent/blob/main/rdagent/scenarios/rl/autorl_bench/README.md)\n  Benchmark for agents that autonomously design, implement, debug, and execute RL post-training pipelines.\n\n- **Ambig-DS: A Benchmark for Task-Framing Ambiguity in Data-Science Agents** (2026) - [Paper](https://arxiv.org/abs/2605.09698) | [Data](https://huggingface.co/datasets/anonymous222bit/Ambig-DS-T)\n  Diagnostic suites for silent target and objective misframing in data-science agents, built on DSBench and MLE-bench.\n\n- **DSGym: A Holistic Framework for Evaluating and Training Data Science Agents** (2026) - [Paper](https://arxiv.org/abs/2601.16344) | [Code](https://github.com/fannie1208/DSGym)\n  Modular execution framework and task suite for evaluating and training data science agents in self-contained environments.\n\n- **FML-bench: A Controlled Study of AI Research Agent Strategies from the Perspective of Search Dynamics** (2026) - [Paper](https://arxiv.org/abs/2605.17373) | [Code](https://github.com/qrzou/FML-bench)\n  Benchmark with 18 fundamental ML research tasks and process-level metrics for studying agent search dynamics.\n\n### Autonomous Data Science Agents\n\n*LLM agents for end-to-end data science automation with case-based reasoning and knowledge retrieval.*\n\n- **DS-Agent: Automated Data Science by Empowering LLMs with Case-Based Reasoning** (ICML 2024) - [Paper](https://arxiv.org/abs/2402.17453) | [Code](https://github.com/guosyjlu/DS-Agent)\n  Two-stage framework (development + deployment) using case-based reasoning from Kaggle. GPT-4 achieves 100% success rate in development stage.\n\n- **Agent Laboratory: Using LLM Agents as Research Assistants** (2025) - [Paper](https://arxiv.org/abs/2501.04227) | [Code](https://github.com/SamuelSchmidgall/AgentLaboratory)\n  End-to-end autonomous research framework with specialized agents (PhD, Postdoc, ML Engineer, Professor) for literature review, experimentation, and report writing. 84% cost reduction vs. prior methods.\n\n- **DataMaster: Data-Centric Autonomous AI Research** (2026) - [Paper](https://arxiv.org/abs/2605.10906)\n  Data-agent framework using tree-structured search, shared data pools, and cumulative memory to optimize datasets through downstream feedback.\n\n### Multi-Agent Systems\n\n*Frameworks using multiple specialized agents for end-to-end ML pipelines.*\n\n- **AIBuildAI: An AI Agent for Automatically Building AI Models** (2026) - [Paper](https://arxiv.org/abs/2604.14455) | [Code](https://github.com/aibuildai/AI-Build-AI)\n  Hierarchical manager, designer, coder, and tuner agents for end-to-end AI model development on MLE-bench tasks.\n\n- **A Multi-Agent Framework for Code-Guided, Modular, and Verifiable Automated Machine Learning** (2026) - [Paper](https://arxiv.org/abs/2602.13937)\n  Introduces iML with code-guided planning, modular preprocessing/modeling components, and contract-based self-correction.\n\n- **AutoML-Agent: A Multi-Agent LLM Framework for Full-Pipeline AutoML** (ICML 2025) - [Paper](https://openreview.net/forum?id=p1UBWkOvZm) | [Code](https://github.com/DeepAuto-AI/automl-agent)  \n  Multi-agent system with data, model, and operation agents for full-pipeline automation.\n\n- **AutoAgents: Generating Specialized Agents for End-to-End Machine Learning Pipelines** (2024) - [Paper](https://arxiv.org/abs/2410.03521)\n  Dynamically composes specialized agents across data prep, training, and evaluation stages, with reported AutoML task speedups.\n\n- **LightAutoDS-Tab: Multi-AutoML Agentic System for Tabular Data** (2025) - [Paper](https://arxiv.org/abs/2507.13413) | [Code](https://github.com/sb-ai-lab/LADS)  \n  Combines LLM-based code generation with multiple AutoML tools (AutoGluon, LightAutoML, FEDOT).\n\n- **MLZero: A Multi-Agent System for End-to-end Machine Learning Automation** (NeurIPS 2025) - [Paper](https://arxiv.org/abs/2505.13941) | [Code](https://github.com/autogluon/autogluon-assistant)  \n  Transforms raw multimodal data into ML solutions with zero human intervention.\n\n- **SmartDS-Solver: Agentic AI for Vertical Domain Problem Solving in Data Science** (ICLR 2026 Submission) - [Paper](https://openreview.net/forum?id=r7gmePFADZ)\n  Reasoning-centric system with SARTE algorithm for data science problem solving.\n\n- **AutoKaggle: A Multi-Agent Framework for Autonomous Data Science Competitions** (2024) - [Paper](https://arxiv.org/abs/2410.20424)\n  Multi-agent framework for autonomous Kaggle competitions with 85% valid submission rate. Iterative development with code execution, debugging, and unit testing.\n\n- **The AI Data Scientist** (2025) - [Paper](https://arxiv.org/abs/2508.18113)\n  Multi-subagent architecture with six specialized subagents (Data Cleaning, Hypothesis, Preprocessing, Feature Engineering, Model Training, Call-to-Action) for end-to-end workflows.\n\n- **Data Interpreter: An LLM Agent for Data Science** (ICLR 2025) - [Paper](https://arxiv.org/abs/2402.18679) | [Code](https://github.com/geekan/MetaGPT)\n  Hierarchical graph modeling with programmable node generation. Outperforms AutoGen on ML benchmarks.\n\n- **MetaAgent: Automatically Constructing Multi-Agent Systems Based on Finite State Machines** (ICML 2025) - [Paper](https://arxiv.org/abs/2507.22606)\n  FSM-based framework that auto-generates multi-agent systems with state traceback for self-correction. Matches or exceeds human-designed systems.\n\n- **Toward Autonomous Long-Horizon Engineering for ML Research** (2026) - [Paper](https://arxiv.org/abs/2604.13018) | [Code](https://github.com/AweAI-Team/AiScientist)\n  AiScientist coordinates specialist agents through durable workspace artifacts for paper reproduction and MLE competition workflows.\n\n### Search \u0026 Planning Methods\n\n*Papers using tree search, MCTS, or structured planning for ML workflow optimization.*\n\n- **AI Research Agents for Machine Learning** (2025) - [Paper](https://arxiv.org/abs/2507.02554) | [Code](https://github.com/facebookresearch/aira-dojo)  \n  Formalizes AI research agents as search policies with operators. Compares Greedy, MCTS, and Evolutionary strategies.\n\n- **AIRA_2: Overcoming Bottlenecks in AI Research Agents** (2026) - [Paper](https://arxiv.org/abs/2603.26499)\n  Uses asynchronous multi-GPU workers, hidden consistent evaluation, and dynamic ReAct operators to scale MLE-bench-30 search.\n\n- **AutoMind: Adaptive Knowledgeable Agent for Automated Data Science** (2025) - [Paper](https://arxiv.org/abs/2506.10974) | [Code](https://github.com/zjunlp/AutoMind)  \n  Features curated expert knowledge base from 455 Kaggle competitions, agentic knowledgeable tree search, and self-adaptive coding strategy.\n\n- **Can We Predict Before Executing Machine Learning Agents?** (ACL 2026) - [Paper](https://arxiv.org/abs/2601.05930) | [Code](https://github.com/zjunlp/predict-before-execute)\n  FOREAGENT uses a Predict-then-Verify loop to prune expensive ML execution, accelerating convergence while improving outcomes.\n\n- **I-MCTS: Enhancing Agentic AutoML via Introspective Monte Carlo Tree Search** (2025) - [Paper](https://arxiv.org/abs/2502.14693) | [Code](https://github.com/jokieleung/I-MCTS)  \n  Introspective node expansion with hybrid LLM-estimated and actual performance rewards.\n\n- **MLE-STAR: Machine Learning Engineering Agent via Search and Targeted Refinement** (2025) - [Paper](https://arxiv.org/abs/2506.15692) | [Blog](https://research.google/blog/mle-star-a-state-of-the-art-machine-learning-engineering-agents/)  \n  Uses web search to retrieve models and targeted code block refinement via ablation studies.\n\n- **ML-Master: Towards AI-for-AI via Integration of Exploration and Reasoning** (2025) - [Paper](https://arxiv.org/abs/2506.16499) | [Code](https://github.com/sjtu-sai-agents/ML-Master)\n  Integrates exploration and reasoning with adaptive memory mechanism.\n\n- **ML-Master 2.0: Toward Ultra-Long-Horizon Agentic Science** (2026) - [Paper](https://arxiv.org/abs/2601.10402) | [Code](https://github.com/sjtu-sai-agents/ML-Master)\n  Hierarchical Cognitive Caching (HCC) for ultra-long-horizon ML engineering. Achieves 56.44% on MLE-bench via cognitive accumulation across 24-hour runs.\n\n- **PiML: Automated Machine Learning Workflow Optimization using LLM Agents** (AutoML 2025) - [Paper](https://openreview.net/forum?id=Nw1qBpsjZz)\n  Persistent iterative framework with adaptive memory and systematic debugging.\n\n- **Reasoning as Gradient: Scaling MLE Agents Beyond Tree Search** (2026) - [Paper](https://arxiv.org/abs/2603.01692) | [Code](https://github.com/microsoft/RD-Agent)\n  Introduces Gome, which maps diagnostic reasoning, success memory, and multi-trace execution into a directed optimization process for MLE agents.\n\n- **RF-Agent: Automated Reward Function Design via Language Agent Tree Search** (2026) - [Paper](https://arxiv.org/abs/2602.23876) | [Code](https://github.com/deng-ai-lab/RF-Agent)\n  Frames reward-function design as sequential decision-making with MCTS-guided language agents, showing gains across 17 low-level control tasks.\n\n- **SELA: Tree-Search Enhanced LLM Agents for Automated Machine Learning** (2024) - [Paper](https://arxiv.org/abs/2410.17238) | [Code](https://github.com/geekan/MetaGPT/tree/main/metagpt/ext/sela)  \n  Leverages MCTS to expand the search space with insight pools.\n\n### Domain-Specific Agentic ML\n\n*Agentic systems tailored for specific ML domains.*\n\n- **AlphaEvolve: A Coding Agent for Scientific and Algorithmic Discovery** (2025) - [Paper](https://arxiv.org/abs/2506.13131) | [Blog](https://deepmind.google/blog/alphaevolve-a-gemini-powered-coding-agent-for-designing-advanced-algorithms/)\n  Google DeepMind's evolutionary coding agent pairing Gemini LLMs with automated evaluators. Discovered novel matrix multiplication algorithms and optimizes data center operations.\n\n- **AgenticSciML: Collaborative Multi-Agent Systems for Emergent Discovery in Scientific ML** (2025) - [Paper](https://arxiv.org/abs/2511.07262)\n  Specialized agents propose, critique, and refine SciML solutions.\n\n- **AI-Driven Automation Can Become the Foundation of Next-Era Science of Science Research** (NeurIPS 2025 Position) - [Paper](https://openreview.net/forum?id=u0FB996GIH)  \n  Position paper on AI automation for scientific discovery with multi-agent systems to simulate research societies.\n\n- **ClimateAgent: Multi-Agent Orchestration for Complex Climate Data Science Workflows** (TMLR) - [Paper](https://openreview.net/forum?id=XLWvXNumGa)  \n  Multi-agent framework for end-to-end climate data analytics with dynamic API awareness and self-correction.\n\n- **The AI Cosmologist: Agentic System for Automated Data Analysis** (2025) - [Paper](https://arxiv.org/abs/2504.03424)  \n  Automates cosmological data analysis from idea generation to research dissemination.\n\n- **TS-Agent: Structured Agentic Workflows for Financial Time-Series Modeling** (2025) - [Paper](https://arxiv.org/abs/2508.13915)  \n  Modular framework for financial forecasting with structured knowledge banks.\n\n### LLM-Based ML Optimization\n\n*Using LLMs for specific ML optimization tasks.*\n\n- **Using Large Language Models for Hyperparameter Optimization** (2023) - [Paper](https://arxiv.org/abs/2312.04528)\n  Iterative HPO via LLM prompting. Matches or outperforms Bayesian optimization in limited-budget settings.\n\n- **ML-Agent: Reinforcing LLM Agents for Autonomous Machine Learning Engineering** (2025) - [Paper](https://arxiv.org/abs/2505.23723)\n  Applies online reinforcement learning to train LLM agents for ML tasks with exploration-enriched fine-tuning and step-wise RL.\n\n- **AceGRPO: Adaptive Curriculum Enhanced Group Relative Policy Optimization for Autonomous Machine Learning Engineering** (2026) - [Paper](https://arxiv.org/abs/2602.07906) | [Code](https://github.com/yuzhu-cai/AceGRPO)\n  Reinforcement learning method for MLE agents using evolving execution traces and learnability-guided adaptive sampling.\n\n- **Synthetic Sandbox for Training Machine Learning Engineering Agents** (2026) - [Paper](https://arxiv.org/abs/2604.04872)\n  SandMLE generates micro-scale synthetic MLE environments to make on-policy trajectory-wise RL practical for ML engineering agents.\n\n- **CAAFE: Context-Aware Automated Feature Engineering** (NeurIPS 2024) - [Paper](https://arxiv.org/abs/2305.03403)\n  LLM-driven automated feature engineering pipeline that generates and executes code for new features using dataset context.\n\n- **LLM-FE: Automated Feature Engineering for Tabular Data with LLMs as Evolutionary Optimizers** (2025) - [Paper](https://arxiv.org/abs/2503.14434) | [Code](https://github.com/nikhilsab/LLMFE)\n  Combines evolutionary search with LLM reasoning to discover effective feature transformations. Outperforms CAAFE and other baselines on classification and regression benchmarks.\n\n- **A Human-Centered Automated Machine Learning Agent with LLMs** (2025) - [Paper](https://www.frontiersin.org/journals/artificial-intelligence/articles/10.3389/frai.2025.1680845/full)\n  LLM-driven agent enabling natural language interaction throughout the entire ML workflow with adaptive hyperparameter optimization.\n\n### Surveys\n\n*Survey papers covering the agentic ML landscape.*\n\n- **Large Language Models for Constructing and Optimizing Machine Learning Workflows: A Survey** (ACM TOSEM, 2025) - [Paper](https://dl.acm.org/doi/10.1145/3773084)\n  First SE-oriented, stage-wise review of LLM-based ML workflow automation covering data/feature engineering, model selection, HPO, and evaluation.\n\n- **Large Language Model-based Data Science Agent: A Survey** (2025) - [Paper](https://arxiv.org/abs/2508.02744)\n  Comprehensive survey examining how LLM-agent systems automate end-to-end data science pipelines.\n\n- **A Survey on Large Language Model-based Agents for Statistics and Data Science** (The American Statistician, 2025) - [Paper](https://www.tandfonline.com/doi/full/10.1080/00031305.2025.2561140)\n  Overview of evolution, capabilities, and applications of LLM-based data agents for simplifying complex data tasks.\n\n- **Agentic AI for Scientific Discovery: A Survey of Progress, Challenges, and Future Directions** (2025) - [Paper](https://arxiv.org/abs/2503.08979)\n  Categorizes agentic systems for scientific discovery into autonomous and collaborative frameworks across chemistry, biology, and materials science.\n\n- **From AI for Science to Agentic Science: A Survey on Autonomous Scientific Discovery** (2025) - [Paper](https://arxiv.org/abs/2508.14111)\n  Domain-oriented review unifying process-oriented, autonomy-oriented, and mechanism-oriented perspectives on autonomous scientific discovery.\n\n- **From Automation to Autonomy: A Survey on Large Language Models in Scientific Discovery** (2025) - [Paper](https://arxiv.org/abs/2505.13259)\n  Analyzes LLM progression from discrete task-oriented functions to sophisticated multi-stage agentic workflows across six stages of the scientific method.\n\n- **A Comprehensive Survey of Self-Evolving AI Agents** (2025) - [Paper](https://arxiv.org/abs/2508.07407) | [Code](https://github.com/EvoAgentX/Awesome-Self-Evolving-Agents)\n  Survey on agent evolution techniques that automatically enhance agent systems, bridging foundation models with continuous adaptability.\n\n### Foundation Models for ML\n\n*Pre-trained models that enable rapid ML development.*\n\n- **TabPFN: A Transformer That Solves Small Tabular Classification Problems in a Second** (ICLR 2023) - [Paper](https://arxiv.org/abs/2207.01848) | [Code](https://github.com/automl/TabPFN)  \n  Prior-Data Fitted Network using in-context learning for instant tabular classification.\n\n- **Unlocking the Full Potential of Data Science Requires Tabular Foundation Models, Agents, and Humans** (NeurIPS 2025 Position) - [Paper](https://openreview.net/forum?id=aXMPvmBAm5)  \n  Position paper on collaborative systems integrating agents, tabular foundation models, and human experts for data science.\n\n---\n\n## Datasets \u0026 Benchmarks\n\n*Benchmarks and datasets for evaluating agentic ML systems.*\n\n| Benchmark | Description | Link |\n|-----------|-------------|------|\n| Agent^2 RL-Bench | Benchmark for agents that autonomously engineer RL post-training loops for foundation models. | [Paper](https://arxiv.org/abs/2604.10547) \\| [GitHub](https://github.com/microsoft/RD-Agent/blob/main/rdagent/scenarios/rl/autorl_bench/README.md) |\n| Ambig-DS | Diagnostic suites for target and objective ambiguity in data-science agents. | [Paper](https://arxiv.org/abs/2605.09698) \\| [Data](https://huggingface.co/datasets/anonymous222bit/Ambig-DS-T) |\n| AutoML-Agent Benchmark | 18 diverse datasets across tabular, CV, NLP, time-series, and graph tasks. | [Paper](https://openreview.net/forum?id=p1UBWkOvZm) |\n| DataSciBench | Comprehensive data science benchmark with TFC framework for LLM evaluation. | [Paper](https://arxiv.org/abs/2502.13897) \\| [GitHub](https://github.com/THUDM/DataSciBench) |\n| DS-1000 | 1,000 data science code generation problems from StackOverflow across 7 libraries. | [Paper](https://arxiv.org/abs/2211.11501) |\n| DSGym | Modular task suite and execution environment for evaluating and training data science agents. | [Paper](https://arxiv.org/abs/2601.16344) \\| [GitHub](https://github.com/fannie1208/DSGym) |\n| FML-bench | 18 fundamental ML research tasks with process-level metrics for agent search dynamics. | [Paper](https://arxiv.org/abs/2605.17373) \\| [GitHub](https://github.com/qrzou/FML-bench) |\n| GAIA | General AI Assistants benchmark testing real-world reasoning and tool use. | [Paper](https://arxiv.org/abs/2311.12983) |\n| LMR-BENCH | Benchmark tasking agents with reproducing functions from NLP research papers. | [Paper](https://aclanthology.org/2025.emnlp-main.314/) |\n| MLE-bench | Kaggle-based benchmark for ML engineering agents by OpenAI. 75 competitions. | [Paper](https://arxiv.org/abs/2410.07095) \\| [GitHub](https://github.com/openai/mle-bench) |\n| MLE-Dojo | Gym-style RL environment built on MLE-bench with 200+ Kaggle competitions for agent training. | [Paper](https://arxiv.org/abs/2505.07782) |\n| MLE-Smith | Automated pipeline for generating competition-style MLE challenges from raw datasets. | [Paper](https://arxiv.org/abs/2510.07307) |\n| MLAgentBench | Benchmark for LLM agents on ML experimentation tasks. | [Paper](https://openreview.net/forum?id=1Fs1LvjYQW) |\n| MLR-Bench | Open-ended ML research benchmark with 201 tasks from major ML conferences. | [Paper](https://arxiv.org/abs/2505.19955) |\n\n---\n\n## MLE-bench Leaderboard\n\n*Top-performing agents on [MLE-bench](https://github.com/openai/mle-bench) (75 Kaggle competitions, ICLR 2025 Oral). Scored by \"Any Medal %\" — percentage of competitions earning at least a bronze medal. Agents run on 36 vCPUs, 440GB RAM, and one 24GB A10 GPU. Official leaderboard submissions have been paused since April 24, 2026 while OpenAI develops an improved comparability process.*\n\n| Rank | Agent | LLM | All (%) | Hours | Date |\n|------|-------|-----|---------|-------|------|\n| 1 | Famou-Agent 2.0 | Gemini-3-Pro-Preview | 64.44 | 24 | Feb 2026 |\n| 2 | AIBuildAI | Claude-Opus-4.6 | 63.11 | 24 | Mar 2026 |\n| 3 | CAIR MARS+ | Gemini-3-Pro-Preview | 62.67 | 24 | Feb 2026 |\n| 4 | MLEvolve | Gemini-3-Pro-Preview | 61.33 | 12 | Feb 2026 |\n| 5 | PiEvolve (Fractal AI) | Gemini-3-Pro-Preview | 61.33 | 24 | Jan 2026 |\n| 6 | Famou-Agent 2.0 | Gemini-2.5-Pro | 59.56 | 24 | Dec 2025 |\n| 7 | ML-Master 2.0 (SJTU) | DeepSeek-V3.2-Speciale | 56.44 | 24 | Dec 2025 |\n| 8 | CAIR MARS | Gemini-3-Pro-Preview | 56.00 | 24 | Jan 2026 |\n| 9 | PiEvolve (Fractal AI) | Gemini-3-Pro-Preview | 52.00 | 12 | Jan 2026 |\n| 10 | Leeroo | Gemini-3-Pro-Preview | 50.67 | 24 | Dec 2025 |\n| 11 | Thesis | gpt-5-codex | 48.44 | 24 | Nov 2025 |\n| 12 | CAIR MLE-STAR-Pro-1.5 | Gemini-2.5-Pro | 44.00 | 24 | Nov 2025 |\n| 13 | Famou-Agent | Gemini-2.5-Pro | 43.56 | 24 | Oct 2025 |\n| 14 | Operand ensemble | gpt-5 + multi-model | 39.56 | 24 | Oct 2025 |\n| 15 | CAIR MLE-STAR-Pro-1.0 | Gemini-2.5-Pro | 38.67 | 12 | Nov 2025 |\n| 16 | InternAgent | DeepSeek-R1 | 36.44 | 12 | Sep 2025 |\n| 17 | R\u0026D-Agent | gpt-5 | 35.11 | 12 | Sep 2025 |\n| 18 | Neo multi-agent | Undisclosed | 34.22 | 36 | Jul 2025 |\n| 19 | AIRA-dojo (Meta) | o3 | 31.60 | 24 | May 2025 |\n| 20 | R\u0026D-Agent | o3 + GPT-4.1 | 30.22 | 24 | Aug 2025 |\n\n*Top score improved from 16.9% (Oct 2024) to 64.4% (Feb 2026) — a ~3.8x improvement in 16 months. See [MLE-bench README](https://github.com/openai/mle-bench) for the full leaderboard with per-difficulty breakdowns and grading reports.*\n\n---\n\n## Related Resources\n\n*Curated reading lists and paper collections on agentic ML.*\n\n| Resource | Description | Stars |\n|----------|-------------|-------|\n| [LLM4AutoML](https://github.com/t-harden/LLM4AutoML) | Curated list of papers on using LLMs for AutoML. | ![GitHub stars](https://img.shields.io/github/stars/t-harden/LLM4AutoML?style=flat-square) |\n| [LLM-Based Data Science Agent Reading List](https://github.com/Stephen-SMJ/Reading-List-of-Large-Language-Model-Based-Data-Science-Agent) | Reading list of papers on LLM-based data science agents. | ![GitHub stars](https://img.shields.io/github/stars/Stephen-SMJ/Reading-List-of-Large-Language-Model-Based-Data-Science-Agent?style=flat-square) |\n| [ai-agent-papers](https://github.com/masamasa59/ai-agent-papers) | Biweekly-updated collection of AI agent research papers. | ![GitHub stars](https://img.shields.io/github/stars/masamasa59/ai-agent-papers?style=flat-square) |\n| [Awesome-Self-Evolving-Agents](https://github.com/EvoAgentX/Awesome-Self-Evolving-Agents) | Survey and paper list on self-evolving AI agents bridging foundation models and lifelong systems. | ![GitHub stars](https://img.shields.io/github/stars/EvoAgentX/Awesome-Self-Evolving-Agents?style=flat-square) |\n| [Awesome-Agent-Papers](https://github.com/luo-junyu/Awesome-Agent-Papers) | Up-to-date survey on LLM agent methodology, applications, and challenges. | ![GitHub stars](https://img.shields.io/github/stars/luo-junyu/Awesome-Agent-Papers?style=flat-square) |\n\n---\n\n## Contributing\n\nContributions are welcome! To add a project or paper, simply [open an issue](../../issues) or submit a PR.\n\nWhen proposing additions, include a short note on which inclusion criteria the item satisfies and link the strongest supporting evidence (paper/repo/benchmark/blog).\n\n---\n\n## License\n\n[![CC0](https://licensebuttons.net/p/zero/1.0/88x31.png)](https://creativecommons.org/publicdomain/zero/1.0/)\n\nTo the extent possible under law, the authors have waived all copyright and related rights to this work.\n","projects_url":"https://awesome.ecosyste.ms/api/v1/lists/jxucoder%2Fawesome-agentic-machine-learning/projects"}