Staff Software Engineer - Agentic AI Compute

We are looking for experienced and motivated engineers who are up for a new challenge to be part of a team building the compute platform for AI at scale! Job Description Agentic AI is changing what CPU performance means in the cloud . As AI systems evolve from individual model calls towards complex, distributed workflows, performance increasingly depends on efficiently orchestrating models, data, tools and services across the whole system . You will join a diverse, dedicated and inclusive group based in Arm's vibrant office at the heart of Manchester. Your work will help ensure the latest developments in AI take full advantage of Arm Neoverse platforms and pioneering cloud-server technologies like AGI-CPU, our CPU for next-generation AI data centers. Together with teams from across Arm, the open source community, and our partners, you will be accelerating AI and helping build the future of computing on Arm. We work on the full stack, from applications and frameworks, to compilers, libraries, and assembly; we encourage everyone in the team to get involved in all parts. The team has a wide remit, but we are all aligned on a common goal. We trust and empower engineers with the autonomy to identify and solve problems without silos and with the support of their team mates. Key skills We're looking for people who are motivated to learn, we will support your growth. You don't need all of these to be successful! Excellent software engineering and programming skills, preferably C++ and Python Eager to dive into unfamiliar software stacks and codebases , debug complex issues, and identify performance opportunities Familiarity with Linux development environments , profiling, debugging and performance analysis An interest in understanding end-to-end application and system performance , from higher-level applications through runtimes and libraries to the underlying hardware Experience being responsible for the delivery of sophisticated technical work involving multiple contributors A degree in a computational or numerate subject , or equivalent experience Nice-to-have skills and experience We’d also be interested to hear about any experience you have in the following areas: Experience with ML frameworks and LLM inference , such as PyTorch, vLLM, SGLang, llama.cpp or similar Experience analysing and optimizing CPU performance in heterogeneous CPU + accelerator systems Familiarity with agentic AI frameworks and protocols Understanding of CPU bottlenecks in agentic workloads Understanding of databases, vector search and retrieval systems Familiarity with containers and cloud/server infrastructure Familiarity with batching , scheduling , KV-cache management, collective communications and distributed AI workloads Practical understanding of parallel programming Comfortable working across abstractions in the stack, from low-level optimisation to high level operators Experience engaging with the open-source community and contributing to large open-source software projects Experience optimizing software for AArch64/Arm servers , including SIMD technologies such as Neon™, SVE and SME Interpersonal strengths Strong and clear communication skills Able to work independently, make pragmatic decisions and navigate ambiguity and change People management or mentoring experience - wellbeing, performance review, supporting career development and coaching In Return You will receive an attractive reward package with generous stock awards. We also know that people do their best work when they feel rested and psychologically safe, so we have a progressive leave policy including generous parental leave, days of care and 4 weeks of sabbatical every 4 years. We are flexible on working hours, and allow hybrid working. We strive to provide an open and accepting environment where you are encouraged to share your ideas and opinions; which enables collective innovation and creativity, and supports your professional and personal growth. We are an equal opportunity employ