Research Scientist, Hedra

• We’re looking for experienced researchers in Generative AI, especially with a focus on image and video generation
• 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
• Proficient in deep learning frameworks such as PyTorch, Jax, or similar
• Knowledgeable in advanced model training techniques and data management
• Strong research track record with publications in top-tier conferences or journals

• Join our team to work on groundbreaking projects that push the limits of video generation technology and multi-modal control
• 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
• Resources: Access to advanced computing resources and collaboration with top cloud providers

• Equity
• 401k (no match)
• Healthcare (Silver PPO Medical, Vision, Dental)
• Gym access (at our office)

Total funding: $42m

The relationship between machine learning and online content has grown closer in recent years, and Hedra is a video content platform that is pushing this symbiosis into new territory. Describing itself as a video foundation mode company, Hedra is building an online creation studio that helps users generate dialogue, audio and video from words.