AI System Architect

AI System Architect Overview

This opportunity is for an experienced AI system architect to join a UK startup in Oxfordshire that is building an AI accelerator for data centres. The company is developing hardware intended to deliver much faster computation while using significantly less energy than conventional approaches, with the broader aim of changing the economics of AI and the way large-scale AI systems scale globally. The role sits in a highly technical, hardware-focused environment where the product is still being shaped, so the successful candidate will have a direct influence on architecture and product direction from an early stage.

What the role involves

• Helping define the architecture of an AI accelerator designed for data-centre use, with a strong emphasis on speed, efficiency, and model performance.
• Working at the intersection of AI workloads and hardware design to ensure the platform is well suited to training and inference use cases.
• Contributing to decisions that affect the full stack, including instruction set architecture, compiler integration, runtime behaviour, and datapath microarchitecture.
• Using analytical modelling methods to evaluate design options, including roofline analysis, memory bandwidth modelling, and cycle-level performance estimation.
• Applying knowledge of how modern AI models behave in practice, especially in relation to compute intensity and memory access patterns.
• Writing clear technical specifications and using data-backed reasoning to influence architecture choices and team decisions.
• Supporting a culture where hardware is built specifically to maximise the performance of AI models rather than as a generic compute platform.

Technical background and experience sought

• Deep expertise in AI and machine learning, with practical understanding of transformers, convolutional neural networks, recommender systems, and the way these workloads stress compute and memory subsystems.
• Previous experience designing systems or chips for machine learning training or inference.
• Ability to carry out detailed performance modelling and translate those results into architectural trade-offs.
• Comfort working across both hardware and software domains, including ISA design, compiler interfaces, runtime layers, and microarchitectural datapaths.
• Industry experience from AI-chip startups or major AI hardware teams, such as those working on TPU, MTIA, Trainium, Dojo, or similar platforms.
• Familiarity with compiler and ML infrastructure such as MLIR, XLA, TVM, and Triton, along with an understanding of how these tools influence hardware design decisions.
• Strong written communication skills and the ability to persuade through structured documentation and evidence-based analysis.
• An experienced profile suited to a senior architecture role rather than an entry-level position.

Working environment and package

• The role is based in Oxfordshire, England and is described as hybrid working.
• The company is presented as a cutting-edge hardware startup with an AI-first culture.
• The position offers the chance to work on frontier problems in AI silicon design.
• Compensation is described as competitive and includes salary, equity, and benefits.
• Flexible working arrangements are available.
• The role offers high impact from the outset, with the opportunity to shape product direction immediately.

How to apply

Interested candidates are invited to apply now or contact Chris Amison at IC Resources for further information. The listing also provides a direct request-to-call option and contact details for the recruiter.

Degree Requirement: Not Specified

Visa Sponsorship May be