Main focus: Graduates in the software team may end up working in any of the areas described in the other software roles. We encourage anyone with an interest in the machine intelligence space to contact us. Working at Graphcore will be very challenging but allows you to work on both software and hardware technologies shaping the future of this industry.
We are looking for bright, motivated graduates from good universities who demonstrate a strong desire to work in cutting edge technology development and who show good ability to work hands-on developing software.
PhD candidates are welcomed from relevant fields with applied software development skills.
Main focus: Developing and maintaining the infrastructure for internal and public cloud based high performance machine learning compute clusters.
Candidates should have experience in systems adminstration and/or
devops. Candidates who have worked on large scale HPC systems or data
centers are particularly suited to this role.
Main focus: Create and maintain the software and systems that allow software development to proceed professionally and smoothly, including continuous integration, build and test, source control and management and management of the software release process.
Experience in software development infrastructure or development operations is preferred but not essential. Candidates should have experience in a combination of software and systems development. The role involves using a range of technologies, such as python, cmake, buildbot, phabricator, AWS etc. Candidates should have experience that shows flexibility in using and evaluating new technology.
Main focus: The primary goal for this role is to ensure that we have a clear understanding and record of the verification and testing that has been done for each software product release. This will involve establishing and maintaining systems so that test data and coverage are reported and managed. You will need to be fully aware of all the development work going on within the software team and of the tests being written by the developers, to identify and rectify testing issues.
This is a hands on role. Candidates should expect to spend a considerable amount of their time developing testing systems and infrastructure.
Previous experience in software verification is preferred as is good software development/system experience.
Apply to careers@graphcore.aiMain focus: Our IPU platform is designed to be targeted by machine learning frameworks such as Tensorflow, MXNet, Caffe2, CNTK and PyTorch. This role involves understanding the implementation of these frameworks at a deep technical level. This role focuses on the integration of the IPU and Poplar platform into these frameworks.
Ideally candidates will have experience of developing machine learning frameworks, but it is not essential. Strong C++ development ability along with an understanding of how these frameworks target existing technologies such as a GPU would be useful. Developers will be expected to gain a deep understanding of the Poplar framework and the functionality provided by its application libraries, quickly upon joining Graphcore, in order to integrate it into machine learning frameworks. Fundamentally, this role provides an
opportunity to work on adding advanced hardware support to software platforms used by many of the worlds largest companies.
Main focus: This role involves the implementation of machine learning algorithms and their primitives. It is an applied research role, studying how various machine learning algorithms are best implemented on our highly parallel IPU architecture, how to use the machine to its best efficiency for a range of new tasks and algorithms and building the libraries and platform to aid IPU users.
Candidates could come from a range of backgrounds (machine learning research, HPC, DSP algorithm coding, other software development) but must have strong software development skills. Previous experience of implementing software requiring a solid understanding of both the algorithm / application domain and the performance of an underlying hardware system is useful.
Experience in C++ is beneficial but not essential, as long as you have experience in a close language and a willingness to learn. PhDs in relevant areas are welcomed.
Main focus: The host runtime provides the interface between development platforms containing IPU devices and the host machine. This role involves working on the host PCI express driver and the supporting low lever user space libraries to provide device access functionality to the Poplar platform. The runtime also provides access functionality for the production test infrastructure required for building IPU products so requires close interaction with the hardware engineering teams.
A strong background in C and C++ development is required. We are looking for candidates with a good understanding of how hardware platforms are built in particular the interfacing of devices to a host machine via PCI Express and USB. Knowledge of Linux kernel driver development is useful but not essential. Knowledge of modern operating systems and memory management is also beneficial. The position requires the candidate to work very much at thesoftware / hardware boundry of a system.
Main focus: Feedback of application performance data and application debugging is essential functionality for Poplar developers. The role focuses on the design and development of tools, specifically the Poplar debugger and Poplar profiler (both graphical and command line) for the analysis of IPU applications. These custom tools specifically target the hardware support in the IPU designed for analysis of massively parallel graph based applications.
The role will be primarily a C and C++ development position. Previous experience of developing tools for parallel processors would be advantageous along with knowledge of debuggers and performance analysis tools. The role will also require buiding graphical tools to present information to users about how an IPU is being utilized.
Main focus: Working on the on the IPU architecture compiler. Understanding code generation & optimization of C / C++ code to the instruction set of the machine. The architecture compiler and its ability to target the IPU for maximum performance and flexibility, is a fundamental component of the Poplar framework.
Compiler experience is important, although experience in a different but related part of the toolchain, with a desire to move into compiler development, would also be considered. Experience in LLVM and C++ are valued.
Main focus: the setup, management and maintenance of product manufacturing processes that contribute to order fulfilment and outgoing quality.
Activities that relate to this will include
o Manufacturing process specification and management
o Test coverage evaluation and management
o Yield monitoring and improvement
o Process stability and repeatability management
o Device characterisation and performance analysis
Apply to careers@graphcore.ai or contact us for more informationMain focus: Working on tools, libraries and drivers to target our new IPU hardware platform efficiently. The software stack involves a mix of low level coding and drivers, toolchain component (compilers, assemblers, linkers etc.) and libraries to target multiple processors working in parallel.
As the software stack crosses many different domains a good candidate may come from many different backgrounds (toolchain developers, HPC developers, DSP engineers, other application developers).
Fundamentally, this is a coding role so being comfortable with programming and having the skills to cope with new challenges is paramount. C++ experience is ideal but not essential; experience in a close language and a willingness to learn is sufficient. PhDs in relevant areas count as good experience.
Apply to careers@graphcore.ai
