Tech Lead Research Scientist​/Engineer, Neural Graphics and World Models

Tech Lead Research Scientist/Engineer, Neural Graphics and World Models - Tik Tok

Location:

San Jose

Employment Type:

Regular

Job Code:

A138238

Share this listing:

About the Team Join the AI-3D engine team, where we push the boundaries of digital interaction by developing a proprietary next-generation neural engine for games and other interactive content experiences across Tik Tok and future creative platforms.

Our goal is to move beyond traditional graphics pipelines, where geometry, animation, physics, materials, lighting, and rendering are predominantly driven by handcrafted systems. We are exploring an AI-native graphics engine in which learned systems can generate and represent scenes, animate characters, simulate behavior, synthesize visual output, and continuously update interactive worlds in response to user intent and changing context. This engine should enable digital worlds that are persistent, controllable, physically coherent, editable, and responsive.

We are looking for a Tech Lead Research Scientist / Engineer to develop AI models, neural representations, data strategies, and evaluation methods for our neural graphics system.

This is a hands-on research role for candidates who combine current ML expertise with depth in 3D or graphics domain. Depending on your primary focus track, you may define research directions and build AI models, data strategies, evaluation systems, or interactive prototypes across generative modeling, world models, 3D modeling and representations, animation, simulation, and rendering.

Strong candidates combine research judgment with hands-on engineering ability: they can identify an important modeling problem, translate it into concrete research and engineering milestones, run rigorous experiments, and work with engineering teams to turn successful results into interactive, reliable engine capabilities.

• Lead research directions in one or more areas of neural graphics, world models, 3D modeling, animation, simulation, or rendering.
• Develop, train, adapt, and evaluate AI models and representations for interactive graphics and world-generation systems.
• Define data and evaluation strategies that connect model behavior to quality, cont rollability, coherence, and interactive performance.
• Build research prototypes and collaborate with engineering teams to bring successful models into real-time engine workflows.
• Analyze results, identify technical risks, and translate research findings into clear next steps for model, data, and system development.
• Produce reproducible research artifacts, communicate technical direction, and mentor team members.

• PhD/MS, or equivalent research experience in machine learning, computer graphics, robotics, applied mathematics, or a related technical field.
• 5 years of relevant research or industry experience in neural graphics, generative AI, computer graphics, world models, or a closely related domain; doctoral research may count toward this experience.
• Strong, up-to-date knowledge of modern machine learning, with deep expertise in at least one area of model architecture, training, data strategy, evaluation, or scaling and hands-on experience training or adapting generative models such as variational autoencoders, latent tokenizers, diffusion or flow-matching models, diffusion transformers, autoregressive models, or multimodal transformers.
• Engineering fluency with ML frameworks such as PyTorch or JAX, with practical understanding of model development, representation learning, model scaling, distributed training, and pre-training and post-training tradeoffs.
• Deep expertise in at least one of the following AI-for-graphics areas: 3D modelling, asset generation, or 3D representations; animation, motion generation, or character behaviour; simulation, physical AI, or neural simulation; rendering or neural rendering; world models or video prediction; or graphics and game-engine workflows.
• Demonstrated ability to formulate original hypotheses, design controlled experiments, and connect data composition, model behavior, and evaluation results to technical decisions.
• Ability to collaborate…