Summary:
Hedra is a pioneering video foundation model company founded by former Stanford PhDs and backed by top investors at Index, A16Z, and Abstract Ventures. We're building a full-stack creation studio capable of control, emotion, and creative intelligence.
We raised 10 million dollars in our seed round to revolutionize content creation. Our flagship product, the Character foundation model, launched in late June and garnered national attention, landing us on the top of Reddit (multiple times!) and Twitter. In early October we released Character-2, which has already produced highly creative and viral content.
Responsibilities:
-
Model Training: Develop and train generative AI models for video, image, text, and audio, focusing on creating high-quality, controllable avatars.
-
Conditioning Methods: Implement and enhance techniques to add controllability to large models, improving their flexibility and performance.
-
Distributed Training: Manage training processes across hundreds or thousands of GPUs using tools like Ray or PyTorch Distributed, ensuring efficient scaling and handling of large datasets.
Competencies:
-
Research Experience:
-
Proven expertise in training generative models, especially for video and image generation.
-
Skilled in applying conditioning methods to improve model control.
-
Experience with large-scale distributed training on extensive GPU clusters.
-
-
Technical Skills:
-
Proficient in deep learning frameworks such as PyTorch, Jax, or similar.
-
Knowledgeable in advanced model training techniques and data management.
-
-
Educational Background:
-
Strong research track record with publications in top-tier conferences or journals.
-
-
Additional Information:
-
Location: Primarily onsite; remote work may be considered for exceptional candidates.
-
Resources: Access to advanced computing resources and collaboration with top cloud providers.
-
Join Hedra and contribute to the forefront of video generation technology and generative AI research.
Benefits:
-
Competitive compensation and equity.
-
401k (no match).
-
Healthcare (Silver PPO Medical, Vision, Dental).
-
Gym access (at our office).
Top Skills
What We Do
Hedra's flagship foundation model, Character-2, allows users to turn any image into an expressive talking character video. Since launching in July, our platform has attracted over a million users and has created countless pieces of viral content. And we're backed by top investors at Index, A16Z, and Abstract Ventures.
Why Work With Us
Hedra released the first ever audio-conditioned video foundation model, which enables the generation of the most realistic and expressive character videos with AI. Looking forward, our team is creating a full platform for the next generation of media creation, built with AI and creative intelligence at its core.
Hedra Offices
OnSite Workspace
Hedra's main office is in San Francisco and secondary hub is in New York.