Software Engineer, Model Inference, DeepMind

Overview

In this role, you will help bring advanced ML research into production-scale systems. You collaborate with researchers and engineers to design efficient serving infrastructures and optimize LLM deployment on production hardware. You will identify bottlenecks, automate tasks, and improve release velocity across ML pipelines. You shape performance by profiling on GPUs/TPUs, tuning frameworks, and implementing scalable inference backends. This opportunity combines cutting-edge AI research with practical engineering to impact real-world applications.

Responsibilities

• Collaborate with Research teams to design production-ready modeling approaches
• Work with infrastructure teams to deliver efficient serving infrastructure
• Identify opportunities to automate tasks, reduce redundancies, and improve model release velocity
• Develop deep understanding of serving frameworks, pre-processing, caching, and related tech
• Profile and optimize ML workloads on hardware accelerators (GPUs/TPUs) and identify bottlenecks across frameworks, compilers (XLA), and custom kernels
• Contribute to building and optimizing model serving infrastructure or inference backends
• Engage in end-to-end LLM deployment lifecycle, from research collaboration to production deployment

• Bachelors degree or equivalent practical experience
• 8 years of software development experience
• 2 years of deploying and maintaining ML models in production
• Experience profiling, configuring, or running ML workloads on hardware accelerators (GPUs/TPUs)
• Experience designing/building/optimizing model serving infrastructure or inference backends

• collaboration
• problem-solving
• ability to work across cross-functional teams
• JAX
• PyTorch
• CUDA