Software Engineer in Test, Graphcore
More information about location
2+ days a week in office
Advanced AI systems and services
Posted today
Advanced AI systems and services
201-500 employees
Posted today
More information about location
2+ days a week in office
201-500 employees
Company mission
To make machine learning faster, easier and more intelligent.
Role
Who you are
- Strong C++ and Python coding skills and solid Linux fundamentals
- Experience testing numerical/performance-sensitive code
- Understanding of processor architectures and profiling on Linux
- Hands-on with reproducibility/determinism, tolerance design, and benchmarking
- Possess excellent written and oral communication skills, good work ethics, high sense of team-work
Desirable
- Experience benchmarking and optimising algorithmic performance, with deep understanding of memory hierarchy (cache/NUMA), multithreading (synchronisation, work partitioning)
- Experience of testing mixed-precision policies (FP32/BF16/FP16/FP8) and stochastic rounding
- Property-based testing with tensor-shape/dtype/layout generators
- Experience testing HPC kernels, ML runtimes/kernels, or similar
- ABI/API stability and packaging for Linux distros, manylinux, wheels
- Hands-on testing for ML frameworks - custom operators/extensions, seeds/determinism, and debugging dispatcher issues
What the job involves
- As a Software Engineer in Test you will be responsible to safeguard the correctness, reliability, and performance of numerical libraries. You’ll design and own end-to-end test strategies across native C++ components and validating low-level kernels
- This is an exciting opportunity to join an expanding team at Graphcore. Kernel Engineering team is responsible for delivering high performance compute library to help customers gain the maximum performance from AI hardware
- Own existing tests coverage and contribute towards improving it
- Improve correctness of existent microbenchmarks, regression tests, validation techniques
- Mentor colleagues within the team, sharing knowledge and providing guidance where appropriate
- Actively engage in and support Agile ways of working within the team
- Debug issues, resolve bugs and generally improve the quality and functionality of the product
Our take
Traditional hardware, mainly GPUs, used for training these models, comes at a high cost, both financially and environmentally, producing millions of dollars in expenses and hundreds of tons of greenhouse gases. In response, Graphcore is developing accelerator chips designed to expedite AI system training while significantly reducing power consumption.
AI accelerators like those crafted by Graphcore specialise in enhancing specific AI applications, such as artificial neural networks, deep learning, and machine learning. Utilising multicore designs and prioritising low-precision arithmetic or in-memory computing, these accelerators elevate the performance of large AI algorithms, achieving breakthrough results in natural language processing and computer vision.
Graphcore has ambitious plans to creat an AI supercomputer set to debut in 2024, aiming to be one of the most powerful in its class. The company is actively seeking strategic partnerships to bring this vision to fruition, signaling its commitment to shaping the future of AI computing.
Freddie
Company Specialist at Welcome to the Jungle
Insights
Company
Funding (last 2 of 6 rounds)
Dec 2020
$222m
SERIES E
Feb 2020
$150m
SERIES D
Total funding: $682m
Company benefits
- Work-Life Balance: We believe you should have a good work-life balance. Everyone's situation is different, so we offer flexible working arrangements and start and finish times, allowing you to choose when to work from home or come into the office
- Wellbeing: We want to ensure that our employees have the resources and support they need to prioritise their health and wellbeing
- Healthcare: We offer Private Medical Insurance (PMI) to our employees to help them take care of their health and that of their loved ones
- Progression: We prioritise learning and development as an integral part of our culture at Graphcore. Our Talent Development team is dedicated to equipping all our employees with the necessary tools, resources, and support to enhance their skills and reach their full potential in their careers
- Pension: We believe that our employees deserve the best when it comes to planning for their future
- Income Protection: Graphcore understands that unforeseen circumstances can happen, and we want to ensure our employees have the necessary financial security in place
Company HQ
Old City, Bristol, UK
Leadership
Nigel Toon
(CEO)Previously the President and CEO at picoChip Designs (processors for telecoms). As well as the VP and GM of Europe at Altera (semiconductors, merged with Intel).
Simon Knowles
(CTO)Founded two successful semiconductor companies, Element14 and Icera, for a combined value of over $1 billion.
Salary benchmarks
We don't have enough data yet to provide salary benchmarks for this role.
Submit your salary to help other candidates with crowdsourced salary estimates.
Share this job
View 22 more jobs at Graphcore