Multimodal AI Scientist

The role

Our client is building a group-level AI Center of Excellence to set the technical direction for artificial intelligence across its operating companies. This role sits at the centre of its work on multimodal intelligence: models that perceive and reason across vision, language, audio and sensor data, and that operate in the physical world rather than only on text.

Responsibilities

• Model architecture. Design and train multimodal architectures that fuse vision, language, audio and sensor streams into a shared representation.
• World models. Advance work on world models and latent representation learning, including JEPA-style and other self-supervised approaches.
• Reasoning. Build reasoning and planning capability into multimodal systems operating in physical environments.
• Simulation and synthetic data. Create simulation environments and synthetic datasets where real-world data is scarce, costly or unsafe to collect.
• Production. Carry models from research into deployment, including the evaluation, monitoring and failure analysis that requires.
• Technical leadership. Publish, patent, and raise the standard of multimodal practice across the group's businesses.

Experience

• Sustained specialisation in multimodal or representation learning rather than general machine learning breadth
• A record built at a leading technology company, a frontier AI research organisation, or a research-intensive enterprise
• Models taken beyond publication into systems that others depend on
• Large-scale distributed training, and fluency in PyTorch or JAX
• The standing to guide AI teams already established within the group's operating companies

Education

A postgraduate degree in computer science, electrical engineering or a related discipline, from a leading institution in the United States or India. A doctorate is preferred.

Technical areas

• Multimodal fusion and cross-modal alignment
• World models and vision–action models
• Latent representation and self-supervised learning
• Reasoning models
• Speech recognition and sensor analytics
• Physics-aware modelling