Graphcore · posted · checked
Staff Power Integrity Engineer
On-site · BristolStaff+
Who can apply
On-site · Bristol
- “Applicants must have the legal right to work in the UK.”
Location: Bristol, UK
About the job
Shape the power-delivery architecture behind Graphcore’s high-performance AI compute systems, from voltage regulation through package, PCB and silicon.
As Staff Power Integrity Engineer, you will provide technical leadership on Graphcore’s most complex power-delivery challenges. Your analysis and judgement will directly influence architecture and product decisions.
You will define modelling strategies, optimise decoupling and analyse PDN impedance, voltage droop, transient response and regulator interactions. You will turn complex results into clear engineering decisions, requirements and sign-off approaches.
You will lead significant PI issues through resolution and create alignment across silicon, package, board, power, systems and validation teams. You will also develop reusable methods and mentor engineers, raising power-integrity capability across Graphcore.
The team and culture
The Signal and Power Integrity team sits at the centre of Graphcore’s hardware development. We work across disciplines to understand electrical behaviour and guide consequential architecture decisions.
You will own complex PI problems and drive them through to resolution. Decisions are grounded in modelling, first-principles reasoning, hardware evidence and clear assessment of uncertainty.
You will challenge assumptions, communicate risk clearly and help teams make pragmatic trade-offs. Technical influence matters more than reporting lines.
What we’re looking for
Deep practical expertise in power integrity and complex power-delivery network design.
Strong understanding of PDN impedance, transient behaviour, voltage droop, resonance, stability and decoupling across high-current digital systems.
Ability to reason across silicon, package, PCB, voltage regulation and system boundaries.
Strong capability in electromagnetic modelling, circuit simulation, model extraction and selecting appropriate abstraction methods.
Show the full posting
Engineering judgement to assess model validity, uncertainty and sensitivity, then translate analysis into architecture and product decisions.