Oxen.ai – Fast Unstructured Data Version Control
Hi HN! This is Greg, founder of https://oxen.ai. Oxen at it's core is a version control system built from the ground up optimized for machine learning data. Just like other VCS, Oxen can a building block for a lot of workflows. We are a couple ex-IBM Watson engineers who have seen the waves of AI and know the importance of matching the correct data to the correct model deployment in ML workflows. The commands are modeled one-for-one after git, but can handle large unstructured datasets of images, videos, audio, text, as well as tabular data. If you have a few minutes, go to…
In plain words
Oxen.ai is a version control system designed specifically for machine learning data, handling large unstructured datasets including images, videos, audio, and text files. Built by former IBM Watson engineers, it uses Git-like commands but is optimized for data workflows rather than code. Written in Rust with modern hashing algorithms, it can index hundreds of thousands of files quickly, making data collaboration as straightforward as code collaboration in traditional version control systems.
written from the facts on this page · September 2026
From the sources
In the maker’s words, at launch
Hi HN! This is Greg, founder of https://oxen.ai. Oxen at it's core is a version control system built from the ground up optimized for machine learning data. Just like other VCS, Oxen can a building block for a lot of workflows. We are a couple ex-IBM Watson engineers who have seen the waves of AI and know the importance of matching the correct data to the correct model deployment in ML workflows. The commands are modeled one-for-one after git, but can handle large unstructured datasets of images, videos, audio, text, as well as tabular data. If you have a few minutes, go to https://github.com/Oxen-AI/oxen-release#-oxen and check it out! Built in Rust, with modern hashing algorithms and network protocols, it can index hundreds of thousands of images, video, audio, text, in seconds. The vision is to help collaboration on data be as easy as collaboration on code. It still feels like we are in the dark ages of iterating on machine learning datasets, downloading zip files, dumping on s3, or using legacy technology like git-lfs. We would also love to see a community grow around open source datasets, just like open source code. There is a web hub at https://www.oxen.ai where you can sign up for free. Let us know what data you are training your models on today! Running models without the proper dataset infrastructure to iterate on can take the "learn" out of machine learning. With Oxen, we help close that loop.
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 26d ago · cactuscompute.com
