microsoft/RD-Agent

Research and development (R&D) is crucial for the enhancement of industrial productivity, especially in the AI era, where the core aspects of R&D are mainly focused on data and models. We are committed to automating these high-value generic R&D processes through R&D-Agent, which lets AI drive data-driven AI. 🔗https://aka.ms/RD-Agent-Tech-Report

What it solves

R&D-Agent automates the industrial research and development process for data-driven scenarios. It reduces the manual effort required to propose, implement, and optimize models and data features (factors) in fields like quantitative finance, data science, and LLM fine-tuning, transforming the R&D cycle into an autonomous loop.

How it works

The framework uses a dual-component methodology: 'R' (Research) for proposing new ideas and 'D' (Development) for implementing them. It operates as a multi-agent system that can iteratively evolve solutions through self-loops. For example, in quantitative finance, it performs coordinated factor-model co-optimization. It supports various backends via LiteLLM, allowing it to integrate with multiple LLM providers including OpenAI, Azure OpenAI, and DeepSeek.

Who it’s for

  • Quantitative Researchers: To automate the discovery and implementation of trading factors and strategies.
  • ML Engineers: To automate model tuning, feature engineering, and benchmark-driven domain adaptation.
  • Data Scientists: To automate the development of predictive models for competitions (e.g., Kaggle).
  • AI Researchers: To automate the extraction of model structures from research papers and reports.

Highlights

  • Top-performing MLE Agent: Currently leads the MLE-bench benchmark for machine learning engineering tasks.
  • Data-Centric Quant Framework: The first multi-agent framework for automating full-stack quantitative strategy R&D.
  • Diverse Scenarios: Supports quantitative trading, data mining, research copilot (paper-to-code), and autonomous LLM fine-tuning (FT-Agent).
  • Extensible Backend: Integrated with LiteLLM for flexible LLM provider switching.

Related

  • Project
    usail-hkust/LLM-MM-Agent

    An AI agent system that automates the end-to-end mathematical modeling process, from problem analysis and model selection to code execution and academic report generation.

  • Project
    Optim-Agent/optim-agent

    An agentic system optimization framework that uses coding agents to automate parameter tuning by reasoning over the semantic meaning of parameters and trial history.

  • Project
    TauricResearch/TradingAgents

    A multi-agent LLM framework that simulates a professional trading firm's workflow to analyze markets and make informed trading decisions.

  • Project
    openml/automlbenchmark

    A framework for evaluating and comparing open-source AutoML systems using standardized environments and curated datasets to ensure reproducible research.