Applied Research Engineer – Dubbing

Synthesia, the leading AI video platform for business, is seeking an Applied Research Engineer to join its Post-Training Team in London. In this role, you will work on cutting-edge generative AI challenges, focusing on applying diffusion models to production video generation, dubbing, and lip-sync. You will implement post-training optimization techniques such as DPO, quantization, pruning, and distillation to enhance model efficiency and avatar quality.

• Apply DPO (direct preference optimisation) to pre-trained models.
• Adapt models to extend capabilities by changing conditioning inputs.
• Build solutions for dubbing and evaluate lip-sync quality.
• Implement post-training optimization techniques including quantization, pruning, and distillation.
• Analyze and address model performance challenges to ensure high-quality avatar rendering.
• Stay current with research in diffusion models, adversarial networks, and post-training methods.

• Bachelor's or higher in Computer Science, Computer Vision, or related field with 3+ years industry experience.
• Knowledge of recent post-training techniques (distillation, adversarial networks, efficient attention).
• Experience with generative models for images/videos (Diffusion/GAN), preferably in the avatar domain.
• Strong research mindset and ability to push boundaries beyond known methods.