AI Researcher (Multimodal Perception Models)

Join Tavus's core AI team to push the frontier of multimodal conversational intelligence. You'll research and develop neural architectures spanning video, audio, and language to enable AI Humans that see, hear, respond, and look genuinely human. Your work will directly shape the future of natural, face-to-face human-machine interaction.

What you'll do

• Research foundational multimodal models for conversational avatars, including neural avatars and talking-head generation.
• Model video, audio, and language sequences using autoregressive, predictive, and diffusion architectures with emphasis on temporal and sequential dynamics.
• Partner with the Applied ML team to translate research breakthroughs into production systems.
• Stay at the cutting edge of multimodal learning and help define what "cutting edge" means next.

What Tavus is looking for

• Advanced research experience in machine learning with strong foundations in deep learning and sequence modeling across video, audio, and language.
• Track record modeling human behavior and generation—such as facial expressions, affect, or speech—in conversational or interactive contexts.
• Deep expertise in modern architectures including Transformers and Diffusion Models, plus familiarity with training large language or vision-language models.
• Excellent programming skills in PyTorch and demonstrated ability to make deep learning systems work in practice.