Research Scientist - Generative AI & Multimodal Models

Google DeepMind; Greater London; Hybrid; GBP 100,000 - 180,000; Full time

Job summary

Google DeepMind in London is seeking a Research Scientist to scale generative AI research and deploy ideas across real-world systems. You'll set up large-scale tests, manage deadlines, and apply the latest theories to build new models and algorithms.

Work spans from designing architectures to publishing findings and collaborating with universities worldwide. You will balance short-term experiments with long-term goals, bridging research and production while striving for safe, high-impact AI.

Qualifications

• PhD in CS/Math/ML or equivalent practical experience.
• Experience training generative models (LLM, image, video).
• Experience with ML frameworks such as TensorFlow, JAX or PyTorch.
• Research track record and publications in AI conferences.

Responsibilities

• Optimize model architectures and training pipelines for distributed training across accelerator clusters.
• Design and train state-of-the-art generative models for multimodal domains (image, video, audio).
• Develop and deploy RL algorithms to align model behaviors and improve multi-step reasoning.
• Improve inference with quantization, pruning, distillation, and other techniques for production.
• Bridge research and production by turning breakthroughs into scalable algorithms.

Skills

Research experience

Generative models training

ML frameworks (TensorFlow, JAX, PyTorh

Education

PhD in computer science, mathematics, applied statistics, ML or equivalent

Tools

TensorFlow

JAX

PyTorch

Job description

Google DeepMind in London is seeking a Research Scientist to scale generative AI research and deploy ideas across real-world systems. You'll set up large-scale tests, manage deadlines, and apply the latest theories to build new models and algorithms.

Work spans from designing architectures to publishing findings and collaborating with universities worldwide. You will balance short-term experiments with long-term goals, bridging research and production while striving for safe, high-impact AI.