Autonomous Data Scientist Agent

6 posts · Jun 2025 – Apr 2026

What this trail covers

Together AI enables the creation of autonomous data scientist agents by integrating open-source LLMs with a secure code execution environment (Together Code Interpreter - TCI). These agents leverage the ReAct (Reasoning and Action) pattern, specifically the CodeAct variant, to perform multi-step data analysis tasks. The system allows for flexible prompt engineering to tune agent behavior and supports dynamic installation of necessary libraries within the sandboxed execution environment. DSGym is used for evaluation.

All Together AI trails

Apr 2026EinsteinArena: Harnessing the collective intelligence of agents in the wild to advance scienceThis post introduces EinsteinArena, a platform for collaborative AI agent research on open scientific problems. It details the platform's architecture, which allows agents to interact, share results, and compete on tasks like mathematical problem-solving. A key achievement highlighted is the discovery of a new lower bound (604) for the Kissing Number problem in dimension 11, which involved multiple agents collaborating to refine an initial construction and overcome numerical precision challenges. The post emphasizes the importance of a live leaderboard, deterministic verifiers, and a collaborative discussion layer for advancing scientific discovery through agentic systems.
Jan 2026DSGym: A holistic framework for evaluating and training data science agentsIntroduces DSGym, a unified framework for evaluating and training data science agents. DSGym standardizes data science benchmarks with a single API and abstractions for datasets, agents, and metrics. It expands benchmark scope with new bioinformatics (DSBio) and Kaggle competition (DSPredict) datasets. The framework includes a data generation pipeline for agent training, demonstrating its effectiveness by training a 4B model on 2,000 generated examples to achieve state-of-the-art performance. The post also highlights failure modes in current LLMs for data science tasks, particularly domain-grounding errors in scientific analysis.
Aug 2025How Together AI Uses AI Agents to Automate Complex Engineering Tasks: Lessons from Developing Efficient LLM Inference SystemsThis post details six patterns for building AI agents to automate complex engineering tasks: Infrastructure Patterns (Good Tools, Documentation, Safe Execution) and Behavioral Patterns (Manage Parallel Sessions, Manage Wait Time, Progress Monitoring). It provides a case study on using these patterns to automate the training pipeline for speculator models, a key technique for accelerating LLM inference. The post highlights how these agents reduce manual intervention, improve consistency and reliability, and decrease turnaround time for developing efficient LLM inference systems.
Jul 2025Back to The Future: Evaluating AI Agents on Predicting Future EventsThis post introduces FutureBench, a new benchmark for evaluating AI agents' ability to predict future events. It details the methodology for generating prediction questions from news and prediction markets, the technical stack used (DeepSeek-V3, Firecrawl, Tavily), and a three-level evaluation framework (framework comparison, tool performance, model capabilities). The post also presents initial results comparing different LLMs (GPT-4.1, Claude3.7, DeepSeek-V3) and analyzes their action patterns and prediction strategies.
Jul 2025DeepSWE: Training a Fully Open-sourced, State-of-the-Art Coding Agent by Scaling RLIntroduces DeepSWE-Preview, a state-of-the-art open-source coding agent trained from Qwen3-32B using only reinforcement learning (RL). Details the training recipe, including scalable dataset curation with R2E-Gym, environment setup with tools like Bash execution and file editing, and a sparse outcome reward model. The training process utilizes an enhanced GRPO++ algorithm with features like Clip High, No KL Loss, and Compact Filtering for stable and performant training of multi-turn agents. Integrates Kubernetes for scalable agent rollout collection, managing thousands of containers for RL training.
Advertisement
Jun 2025From Zero to One: Building An Autonomous and Open Data Scientist Agent from ScratchThis post details the end-to-end implementation of an autonomous data scientist agent using Together's open-source models and the Together Code Interpreter (TCI). It follows the ReAct pattern with CodeAct, where the agent generates Python code for actions. The post highlights the safety and modularity provided by TCI, the flexibility of prompt engineering for tuning agent behavior, and the ability to dynamically install libraries. It also presents a CLI tool for triggering the agent and showcases an example of molecular solubility prediction.