AI Researcher (Multimodal Audio/Video Generation) at Series B multimodal AI lab

About the role

Lead research on audio-visual avatar generation at a cutting-edge lab building real-time conversational humans. You will design diffusion-based models for high-fidelity talking heads and neural avatars, bridging the gap between verbal and non-verbal communication. Partner with engineering teams to ship groundbreaking research into production-ready applications across healthcare and education.

Why this role is remarkable

• Shape the future of human-computer interaction by building AI avatars that see, hear, and respond with human-like emotional intelligence.
• Join a well-funded Series B startup backed by top-tier global venture capital firms with a culture of shipping research directly to production.
• Work directly with founders and a high-caliber research team to define the architecture for real-time, multimodal conversational timing and rendering.

What You Will Do

• Design and develop state-of-the-art diffusion models for long-video generation and audio-visual modeling to create life-like talking heads.
• Drive innovation in multimodal perception and neural rendering, capturing intricate verbal and non-verbal signals in perfectly synchronized flows.
• Mentor junior researchers and set strategic research directions while publishing impactful work at top-tier venues like CVPR and NeurIPS.

Qualifications

• Holds a PhD in Computer Science or a related field with 2-3+ years of experience applying large-scale generative models in industry.
• Possesses deep expertise in diffusion models, PyTorch, and GPU-optimized workflows with a proven track record of publication at premier AI conferences.
• Demonstrates mastery of multimodal generation spanning video and audio, ideally with experience in 3D graphics or Gaussian splatting techniques.