Skip to content

datachain

Model Serving & MLOps

A Python library that turns cloud‑stored files into versioned, typed datasets you can query at warehouse speed

Python Latest 0.57.0 · 2d ago Security brief →

Features

  • Compute Engine: parallel/distributed Python over remote files with async I/O and checkpoint recovery
  • Dataset DB: Pydantic‑typed schemas, versioning, lineage tracking, sub‑second filtering/join/search
  • Knowledge Base: markdown summaries enriched by LLMs for human‑readable data navigation

Recent releases

View all 45 releases →
No immediate action
0.57.0 New feature

Zarr support

No immediate action
0.56.1 New feature

UUID field in datasets

No immediate action
0.56.0 New feature

Public bucket auto‑detection

No immediate action
0.55.2 Breaking risk

Rename datachain_worker → compute

No immediate action
0.55.1 Bug fix

Deterministic read_storage hash

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

About

Stars
2,778
Forks
144
Language
Python

Install & Platforms

Install via
pip

Beta — feedback welcome: [email protected]