ARXIV:2606.03841 · AUTONOMOUS DATA SCIENCE AGENTS · SUBMITTED 03 JUN · 20:32 UTC · FRESHNESS FRESH

VerifiedSource: PDF linkedVerifiedPaperPack: citation fields availablePartialProof: unverified proof status

EvoDS: Self-Evolving Autonomous Data Science Agent with Skill Learning and Context Management

Zherui Yang · Fan Liu · Yansong Ning · Hao Liu · arXiv

Create an autonomous data science agent that evolves its skills and manages contexts for efficient, adaptive machine learning tasks.

Ship in 2-4 weeks›Score8.0Evidence unverified

Opportunity summary

Pain Create an autonomous data science agent that evolves its skills and manages contexts for efficient, adaptive machine learning tasks.

Evidence 0 refs | 4 sources | 83% coverage

Blocker Evidence unverified

Open Build Read PDF Signal Canvas Track

PROBLEM

Create an autonomous data science agent that evolves its skills and manages contexts for efficient, adaptive machine learning tasks. However, existing approaches remain fundamentally limited by their static action sets and lack of principled…

METHOD

Full abstract

Recent progress in Large Language Model (LLM) agents has enabled promising advances in automated data science. However, existing approaches remain fundamentally limited by their static action sets and lack of principled long-horizon context management, hindering their ability to accumulate reusable experience across tasks and operate reliably in multi-stage, iterative data science pipelines. To address these challenges, we introduce EvoDS, a self-evolving autonomous data science agent that learns to expand its skills and adaptively managing long-term context through agentic reinforcement learning. Specifically, EvoDS introduces two key strategies: (1) Autonomous Skill Acquisition (ASA) mechanism, which enables agents to synthesize, validate, and reuse executable skills; and (2) Adaptive Context Compression (ACC) strategy, which treats context management as a learned control problem rather than passive truncation. These strategies are orchestrated within a two-stage multi-agent training scheme, enabling EvoDS to autonomously improve over time. Theoretically, we prove that EvoDS's hierarchical design reduces tool-selection error, and its optimization objective aligns with an information bottleneck principle, ensuring efficient context use. Empirically, EvoDS outperforms state-of-the-art open-source data science agents by an average of 28.9% across four diverse benchmarks while eliminating out-of-token failures. Our code and data are available at https://github.com/usail-hkust/EvoDS.

RESULT

ScienceToStartup currently rates this 8.0/10 on the public viability pass. Specifically, EvoDS introduces two key strategies: (1) Autonomous Skill Acquisition (ASA) mechanism, which enables agents to synthesize, validate, and reuse executable skills; and (2)…

WHY NOW

Autonomous Data Science Agents moved forward this cycle; last verified June 2026. Public score 8.0/10. Implementation evidence is present through a linked repository.

Continue into Read for claims, analysis, references, and neighboring papers.

Opportunity summary

Score8.0

PainCreate an autonomous data science agent that evolves its skills and manages contexts for efficient, adaptive machine learning tasks.

Evidence0 refs | 4 sources | 83% coverage

Blockerno shell-level blocker reported

Analysis summary

Create an autonomous data science agent that evolves its skills and manages contexts for efficient, adaptive machine learning tasks.

VerifiedSource: PDF linkedVerifiedPaperPack: citation fields availablePartialProof: unverified proof status

Competitive landscape

Create an autonomous data science agent that evolves its skills and manages contexts for efficient, adaptive machine learning tasks.

Segment

Autonomous Data Science Agents

Adoption evidence

Public code linked for build inspection

Commercial read

8.0/10 public viability

Direct

not classified

Adjacent

not classified

Substitute

not classified

Unknown

not classified

{ "contract_version": "paper-r2", "paper_id": "11fcaee2-3310-4faa-a94f-4a28f7941371", "arxiv_id": "2606.03841", "canonical_route": "/paper/evods-self-evolving-autonomous-data-science-agent-with-skill-learning-and-context-management", "active_tab": "synced from current hash by the drawer client", "selected_artifact": "evods-self-evolving-autonomous-data-science-agent-with-skill-learning-and-context-management", "endpoints": { "paper_pack": "/api/v1/paper/evods-self-evolving-autonomous-data-science-agent-with-skill-learning-and-context-management/paper-pack", "build_passport": "/api/v1/paper/evods-self-evolving-autonomous-data-science-agent-with-skill-learning-and-context-management/build-passport", "mcp_resource": "sciencetostartup://surfaces/paper-workspace" } }

{ "surface": "paper", "mode": "paper", "query": "EvoDS: Self-Evolving Autonomous Data Science Agent with Skill Learning and Context Management", "normalized_query": "2606.03841", "route": "/paper/evods-self-evolving-autonomous-data-science-agent-with-skill-learning-and-context-management", "paper_ref": "evods-self-evolving-autonomous-data-science-agent-with-skill-learning-and-context-management", "topic_slug": null, "benchmark_ref": null, "dataset_ref": null }

{ "@context": "https://schema.org", "@graph": [ { "@type": "WebPage", "@id": "https://sciencetostartup.com/paper/evods-self-evolving-autonomous-data-science-agent-with-skill-learning-and-context-management#webpage", "url": "https://sciencetostartup.com/paper/evods-self-evolving-autonomous-data-science-agent-with-skill-learning-and-context-management", "name": "EvoDS: Self-Evolving Autonomous Data Science Agent with Skill Learning and Context Management", "description": "Create an autonomous data science agent that evolves its skills and manages contexts for efficient, adaptive machine learning tasks.", "isPartOf": { "@id": "https://sciencetostartup.com/#website" } }, { "@type": "ScholarlyArticle", "@id": "https://sciencetostartup.com/paper/evods-self-evolving-autonomous-data-science-agent-with-skill-learning-and-context-management#scholarlyArticle", "headline": "EvoDS: Self-Evolving Autonomous Data Science Agent with Skill Learning and Context Management", "description": "Create an autonomous data science agent that evolves its skills and manages contexts for efficient, adaptive machine learning tasks.", "url": "https://sciencetostartup.com/paper/evods-self-evolving-autonomous-data-science-agent-with-skill-learning-and-context-management", "sameAs": "https://arxiv.org/abs/2606.03841", "identifier": { "@type": "PropertyValue", "propertyID": "arXiv", "value": "2606.03841" }, "isAccessibleForFree": true, "isPartOf": { "@id": "https://sciencetostartup.com/#website" }, "datePublished": "2026-06-02T16:20:58.000Z", "author": [ { "@type": "Person", "name": "Zherui Yang" }, { "@type": "Person", "name": "Fan Liu" }, { "@type": "Person", "name": "Yansong Ning" }, { "@type": "Person", "name": "Hao Liu" } ], "codeRepository": "https://github.com/usail-hkust/EvoDS", "additionalProperty": [ { "@type": "PropertyValue", "propertyID": "viabilityScore", "value": 8 }, { "@type": "PropertyValue", "propertyID": "researchDomain", "value": "Autonomous Data Science Agents" }, { "@type": "PropertyValue", "propertyID": "commercialReadiness", "value": "code, repo url" } ] }, { "@type": "SoftwareSourceCode", "@id": "https://sciencetostartup.com/paper/evods-self-evolving-autonomous-data-science-agent-with-skill-learning-and-context-management#software", "name": "EvoDS: Self-Evolving Autonomous Data Science Agent with Skill Learning and Context Management - Source Code", "description": "Create an autonomous data science agent that evolves its skills and manages contexts for efficient, adaptive machine learning tasks.", "codeRepository": "https://github.com/usail-hkust/EvoDS", "url": "https://github.com/usail-hkust/EvoDS" }, { "@type": "BreadcrumbList", "itemListElement": [ { "@type": "ListItem", "position": 1, "name": "Home", "item": "https://sciencetostartup.com" }, { "@type": "ListItem", "position": 2, "name": "Autonomous Data Science Agents", "item": "https://sciencetostartup.com/topics" }, { "@type": "ListItem", "position": 3, "name": "EvoDS: Self-Evolving Autonomous Data Science Agent with Skil", "item": "https://sciencetostartup.com/paper/evods-self-evolving-autonomous-data-science-agent-with-skill-learning-and-context-management" } ] }, { "@type": "FAQPage", "mainEntity": [ { "@type": "Question", "name": "What is the startup potential of \"EvoDS: Self-Evolving Autonomous Data Science Agent with Skil\"?", "acceptedAnswer": { "@type": "Answer", "text": "Create an autonomous data science agent that evolves its skills and manages contexts for efficient, adaptive machine learning tasks." } }, { "@type": "Question", "name": "What products could be built from this research?", "acceptedAnswer": { "@type": "Answer", "text": "Package EvoDS into a software solution that can integrate with existing data pipelines, allowing users to customize data science workflows with minimal manual intervention." } }, { "@type": "Question", "name": "What are the practical use cases?", "acceptedAnswer": { "@type": "Answer", "text": "An AI tool for businesses that need to continuously analyze data and adapt their machine learning models without constant human reprogramming, increasing efficiency and cutting costs." } }, { "@type": "Question", "name": "What industries could this research disrupt?", "acceptedAnswer": { "@type": "Answer", "text": "Replaces manual algorithms tuning and setup in data teams, reducing the need for constant human oversight and improving the dynamic capability of data systems." } } ] } ] }

Competitive landscape

Create an autonomous data science agent that evolves its skills and manages contexts for efficient, adaptive machine learning tasks.

Segment

Autonomous Data Science Agents

Adoption evidence

Public code linked for build inspection

Commercial read

8.0/10 public viability

Direct

not classified

Adjacent

not classified

Substitute

not classified

Unknown

not classified

EvoDS: Self-Evolving Autonomous Data Science Agent with Skill Learning and Context Management

EvoDS: Self-Evolving Autonomous Data Science Agent with Skill Learning and Context Management

Claim map

Constellation map

Competitive landscape

Buzz

PDF

REFERENCES

Related Papers

Subscribe to the weekly brief

Build artifacts

Brief

Experiment plan

Validation checklist

Scientific founder

Translational engineer

Domain operator

GTM lead

Regulatory/clinical advisor

Timeline

Claim map

Constellation map

Competitive landscape

Buzz

PDF

REFERENCES

Related Papers

Subscribe to the weekly brief

Build artifacts

Brief

Experiment plan

Validation checklist

Scientific founder

Translational engineer

Domain operator

GTM lead

Regulatory/clinical advisor

Timeline