I am the CEO of Oxen.ai. My background is in AI/ML and have been training deep neural networks since 2013. I've worked on natural language processing and computer vision as well as early iterations of IBM Watson. Since then I have consulted in the AI/ML space building generative image/audio/video models, real time computer vision models, and large language models.

At Oxen.ai are building a GitHub-like platform for collaborating on and customizing AI models. From testing prompts to fine-tuning models on your own data, Oxen.ai lets you own your model weights and datasets end to end. It is powered by open source, scalable version control tool called "oxen" that is built for the size and scale of data needed for modern day ML workflows. From low level merkle tree optimization, to hashing, to network protocols we make collaboration on large scale data easy.

I started Oxen.ai to solve a pain point I'd felt over and over again and am excited to help others manage the data that powers their AI.

Presentations

23x

How Hollywood is using Open Source models to make High Quality VFX Edits

Imagine you are in the middle of editing a Hollywood film and realize you have the actor in the wrong jacket in post. Instead of trying to do a re-shoot, which may cost hundreds of thousands of dollars, can we use AI to fix in post?

This talk demonstrates a practical workflow for production-quality video editing using open-source AI models for video editing. The workflow combines masked region editing, pose guidance, and LoRA fine-tuning to achieve Hollywood-grade results.

See Presentation