Staff Research Engineer - Video Post Training
Core
Apply post-training techniques to foundational generative video models to create realistic synthetic humans for enterprise skill development.
Role type
Staff Research Engineer (Generative Video)
Builds
Production-ready generative video solutions including dubbing, lip-sync, and avatar generation for 50,000+ businesses.
Domain
Generative AI, Computer Vision, Video Synthesis
Deliverable
production ML models
Required skills
Post-training Diffusion models, Direct Preference Optimization (DPO), Model quantization/pruning/distillation, Adversarial networks, Efficient attention mechanisms, Python, Computer Vision, Generative models (Diffusion/GAN)
Preferred skills
Avatar domain experience, SDLC tools (Git, CI/CD)
Technologies
Diffusion models, GANs, Python, Git
Responsibilities
Apply DPO to pre-trained models, Adapt models for extended capabilities, Build solutions for dubbing and lip-sync evaluation, Implement post-training optimization techniques, Analyze and address model performance challenges, Stay updated with latest research in diffusion models and post-training methods
Seniority
Staff, hands-on IC with research leadership