Software Engineer, Model Inference, DeepMind

Google

Software Engineer, Model Inference, DeepMind Overview

Company Name Google
Job Role Software Engineer, Model Inference, DeepMind
Qualifications Not Specified
Category IT Jobs
Job Type Full Time
Location London

DeepMind, a leading AI research organization under Google, is seeking a highly skilled Software Engineer specializing in Model Inference. This role involves working at the intersection of advanced AI research and large-scale system deployment, with a focus on optimizing the performance and efficiency of machine learning models in production environments.

Overview

As a Software Engineer in this position, you will collaborate with research teams to understand and implement innovative modeling techniques, ensuring they are suitable for real-world deployment. You will work closely with infrastructure teams to develop and refine serving systems that handle large language models and other AI models efficiently, addressing challenges related to speed, scale, and quality. Your work will involve performance profiling, system diagnostics, and hardware-level optimization to eliminate bottlenecks and improve throughput on accelerators such as TPUs and GPUs.

Key Responsibilities

  • Partner with research teams to translate next-generation modeling approaches into production-ready systems, considering deployment constraints and performance requirements.
  • Design and implement scalable serving infrastructure that maximizes throughput and minimizes latency, ensuring robustness and reliability.
  • Automate routine tasks, streamline workflows, and enhance the velocity of model releases through automation and process improvements.
  • Gain deep technical understanding of serving frameworks, data pre-processing pipelines, caching mechanisms, and related technologies.
  • Use hardware profiling tools, roofline analysis, and system diagnostics to identify and resolve performance issues across ML frameworks, compilers like XLA, custom kernels such as Pallas, and hardware accelerators.

Qualifications

  • A bachelorâ??s degree or equivalent practical experience in software engineering or a related discipline.
  • At least 8 years of professional experience in software development, with a focus on scalable system design.
  • Minimum of 2 years deploying and maintaining machine learning models in live production environments.
  • Experience working directly with hardware accelerators (GPUs, TPUs) and optimizing workloads on these platforms.
  • Proficiency in developing and optimizing serving infrastructure, including familiarity with ML frameworks such as JAX or PyTorch, and low-level programming models like CUDA, Pallas, or OpenCL.
  • Strong skills in performance profiling, bottleneck analysis, and system optimization, especially in distributed ML systems using parallelism techniques.
  • Understanding of large language model architectures, inference performance dynamics, and related hardware considerations.

About the Role

This position offers the opportunity to work on pioneering AI projects aimed at building general-purpose learning agents. You will be involved in deploying large language models like Gemini onto Googleâ??s infrastructure, impacting a wide range of applications. The role involves close collaboration with research scientists and engineers, with opportunities for both individual contributor and team lead positions. The work emphasizes safety, ethics, and scientific rigor, aligned with Google DeepMindâ??s mission to advance AI for societal benefit.

Compensation and Benefits

In the US, the salary ranges from $207,000 to $300,000 USD, complemented by a 20% bonus target, equity grants, and comprehensive benefits. Google offers a dynamic work environment with opportunities for professional growth and involvement in cutting-edge AI research.

Additional Information

Google is committed to equal opportunity employment and creating an inclusive workplace. The company does not specify visa sponsorship details in this posting, and applicants should verify their work authorization status independently. For those requiring accommodations during the application process, Google provides support through its accommodations program. English proficiency is required for all roles to facilitate global collaboration.

Note: Google does not accept resumes from recruitment agencies and is not responsible for fees related to unsolicited applications. Equity grants are discretionary and subject to approval by Alphabet Inc., with no guaranteed rights or statutory implications.


Degree Requirement: Not Specified

Visa Sponsorship May be

To apply for this job please visit www.google.com.