Staff Research Engineer - Video Post Training

Synthesia Synthesia · Multimodal · EUROPE · Research and Development

Staff Research Engineer role at Synthesia, a leading AI video platform, focusing on post-training generative video creation. The role involves applying and adapting foundational models, implementing optimization techniques, and ensuring high-quality output for synthetic humans. Requires expertise in diffusion models, computer vision, and Python.

What you'd actually do

  1. Apply DPO (direct preference optimisation) to pre-trained models.
  2. Adapt models to extend their capabilities, for instance, by changing conditioning inputs.
  3. Build solutions for dubbing and evaluate the quality of lip-sync.
  4. Implement post-training optimization techniques, such as quantization, pruning and distillation, to improve the efficiency of diffusion models used in avatar generation.
  5. Analyze and address challenges related to model performance, ensuring high-quality output in avatar rendering.

Skills

Required

  • Computer Vision / Computer Science background
  • 3+ years of industry experience
  • Knowledge of recent advancements in post-training techniques (distillation, adversarial networks, efficient attention)
  • Experience with generative models for images and/or videos (Diffusion/GAN)
  • Experience in using most modern frameworks for machine learning and deep learning
  • Great coding skills in Python
  • Experience with SDLC tools (Git)

Nice to have

  • avatar domain experience
  • CI/CD

What the JD emphasized

  • post-training Diffusion models for generative AI
  • post-training techniques
  • generative models for images and/or videos (Diffusion/GAN) preferably in the avatar domain
  • latest research and advancements in diffusion models, adversarial networks and post-training optimization methods

Other signals

  • Generative AI
  • text-to-video
  • Diffusion models
  • post-training optimization