Hypernatural is a generative video platform that puts the creation of great videos in reach for everyone. We’re hiring a Founding ML Engineer to build the future of video creation with us.
Short form video is on an explosive trajectory that shows no signs of slowing down. Every marketer wants to make more videos, but good video is expensive to produce and often requires highly specialized skills.
Enter Hypernatural. We are on a journey to build the most amazing and easy-to-use generative video editor the world has ever seen. We’re shipping custom fine-tuned generative models across multiple modalities (text, imagery, video) and tackling all the crazy latency and multi-player issues involved.
Our engineering ethos
The app prototype is a Django app with a React frontend. We use Postgres and Redis and Celery. It was the fastest way to get something built that we could show to real people, while still being easy to deploy and monitor.
Our frontend is TypeScript, thankyouverymuch dhh. We’re still using React Router — but we’re open to the idea that maybe it’s time to embrace the Next or Remix ecosystem.
We have tests, we have ci, we have automatic deploys. We use a generated TypeScript API client created by some OpenAPI scaffolding, so it’s easy for .ts to talk to .py. But for the first few months there was just an api.ts file with a bunch of fetch() calls in it (!).
You’ve trained and deployed models from start to finish that real users use, without the help of big-company infrastructure. You have an interest in diffusers and have probably fine-tuned one of these yourself for some purpose.
You have an opinion about how to host and optimize large models in production with various budgets and on various timescales.
You’re not afraid to jump into product code to get what you need, if some piece of data is missing or some integration needs to be done.