Research Engineer - Diffusion Models & Multimodal AI

Google Inc.; Greater London; On-site; GBP 110,000 - 140,000; Full time

Job summary

Google DeepMind seeks a Research Engineer for the Gemini Omni project in London. You will apply research ideas to real-world problems, prototype, train models, and deploy solutions at scale across a variety of domains.

The role emphasizes cutting-edge generative media, diffusion models, and multimodal understanding, with collaboration across research and engineering teams to accelerate high-quality product innovation for billions of users.

Qualifications

• Master's degree in computer science, mathematics, applied statistics, machine learning or equivalent practical experience.
• Experience with TensorFlow or ML frameworks (e.g. JAX or PyTorch).
• Experience working in industry, taking projects from proof-of-concept to implementation.
• Experience with training diffusion models.
• Experience conducting applied research.

Responsibilities

• Apply research ideas to high-impact real world problems through prototyping, dataset curation, model training, performance optimization, and deployment.
• Develop cutting-edge techniques in Generative Media (image, video, and audio).
• Optimize algorithms and models for efficient inference.
• Advance capabilities in Multimodal Understanding.
• Perform comprehensive model optimization to enhance performance and scale.

Skills

TensorFlow

JAX

PyTorch

Diffusion models

Applied research

Education

Master's degree in CS, mathematics, statistics or ML

Tools

Diffusion models

ML frameworks

Data pipelines

Job description

Google DeepMind seeks a Research Engineer for the Gemini Omni project in London. You will apply research ideas to real-world problems, prototype, train models, and deploy solutions at scale across a variety of domains.

The role emphasizes cutting-edge generative media, diffusion models, and multimodal understanding, with collaboration across research and engineering teams to accelerate high-quality product innovation for billions of users.