Salience Labs

Software Architect

Salience Labs Oxford, England, United Kingdom

Computer Hardware Manufacturing · 51-200 employees

20 h ago
Remote software-architect Senior (5-10 yrs) Full-time United Kingdom
Create a free account to apply — email only, no card. You can also save this posting or score it against your profile with AI.

About the role

Develop, test, and document custom kernel code for AI accelerator hardware while benchmarking performance for optimization. Contribute to the design of deployment strategies for OCS technology across various AI and machine learning workloads.

What they look for

C++ Python AI Accelerator Hardware Kernel Development Distributed Systems Shared Memory Architecture CUDA NCCL PyTorch TensorFlow HPC Network Architecture GPU Architecture Performance Benchmarking Machine Learning

Requirements

Requires strong programming skills in C++ and Python with a deep understanding of AI accelerator hardware and distributed shared memory architectures. Candidates should possess a passion for innovative technology and the ability to communicate complex technical concepts effectively.

Full description

About us

Salience Labs is building the optical infrastructure that will power the next generation of AI. Our photonic switching technology delivers the high-speed, ultra-low latency networking fabrics AI datacenters need and can’t get anywhere else. We’re a multi-disciplinary team of physicists, engineers, and machine learning specialists on a mission to remove the bottlenecks holding AI infrastructure back.

The Role

You’ll write the software that puts our OCS technology to work for AI. You’ll develop and benchmark custom kernel code on modern AI accelerator hardware, and help define how AI workloads are deployed across training, inference, and agentic use cases. You’ll follow the latest techniques in distributed AI systems and help shape the networks and network devices that come next.

Responsibilities

  • Develop, test, and document custom kernel code for contemporary AI accelerator hardware
  • Develop and run benchmark tests for kernel code, document the results, and derive optimization strategies
  • Contribute to the definition and implementation of our AI workload deployment ecosystem
  • Track the latest techniques for writing and deploying communication kernels in distributed AI accelerator systems
  • Contribute to designing deployment strategies for our OCS technology, tailored to each AI/ML workload type (training, inference, agentic)
  • Play a vital role in the definition and design space exploration of novel networks and network devices

What We’re Looking For

  • Very good programming skills in one or more higher-level languages, with strong experience in C++ and Python
  • A deep understanding of AI accelerator hardware programming models
  • Good knowledge of distributed shared memory architecture concepts
  • You like being part of a team and being helpful to others
  • Curiosity, innovative thinking, and a passion for pushing the boundaries of technology
  • Good written and verbal communication, with the ability to explain complex technical concepts to broad audiences

Desirable

  • Knowledge of AI/ML and/or HPC workloads and the communication patterns they create
  • Knowledge of CUDA and/or NCCL kernels and how they operate on hardware
  • Familiarity with ML frameworks such as PyTorch or TensorFlow and their internal architecture
  • Familiarity with deploying multi-node compute tasks on compute clusters
  • Detailed knowledge of hardware and/or software systems for high performance networking, with GPU architecture and programming models a particular advantage

Similar roles