MLE-agent
VerifiedAI agent that automates ML baselines, debugging, and research workflows.
What is MLE-agent?
MLE-Agent is an open-source CLI companion built for ML engineers and researchers who need reliable assistance with repetitive engineering work. It autonomously sets up baselines, pulls relevant methods from arXiv and Papers with Code, and manages file structures without constant supervision.
The agent works through interactive chat sessions and specialized commands that trigger code generation, automatic debugging between coder and debugger roles, and report synthesis from git history. It also supports Kaggle competition flows and MLOps tool connections for smoother project handoff.
Designed for practitioners who already understand ML concepts, it reduces boilerplate setup and documentation overhead while still requiring user oversight for final decisions and data handling.
What you can build with MLE-agent
Prototype an ML Baseline
Describe a prediction goal in natural language and let the agent generate, train, and locally test an initial model pipeline.
Enter Kaggle Competitions
Run an automated end-to-end mode that prepares submissions with minimal manual steps after the initial project setup.
Produce Weekly Reports
Generate structured progress summaries from local git repositories or via a lightweight web interface connected to GitHub.
Install MLE-agent
pip install -U mle-agent# With pip:
pip install -U mle-agent
# With uv:
uv pip install -U mle-agent- 1Install via pip with 'pip install -U mle-agent' or use uv for faster resolution.
- 2Clone the repository if installing from source and create a virtual environment.
- 3Run 'mle new <project>' to scaffold a new working directory.
- 4Navigate into the project folder and execute 'mle start' to begin the agent session.
- 5Use 'mle chat' for ongoing interactive assistance or 'mle report' for summaries.
MLE-agent: pros & cons
Pros
- +Strong autonomous baseline and debugging capabilities reduce manual coding.
- +Built-in access to research papers and Kaggle workflows in one interface.
- +Generates weekly reports directly from git activity with little setup.
- +Supports multiple LLMs and offers both CLI and optional web modes.
Cons
- –Requires an existing project folder and proper environment configuration before use.
- –Performance depends heavily on the quality of the chosen underlying LLM.
- –Advanced MLOps integrations may need additional manual configuration.
Frequently asked questions
Yes, you must supply keys for the LLM provider you choose to power the agent.
User reviews
Verified reviews from the community shape this listing's rating.
Loading reviews…
Similar agents
Other general-purpose options worth comparing.
