Science jobs in Newcastle upon Tyne

Applied Scientist — Memory & Optimisation

Which experiment would prove you wrong?

In this role, you will turn open questions into precise problems and design the experiments that test them, including the experiment that could prove your favourite idea wrong. The problems sit in memory and optimisation, and in what they cost in compute. We take ideas from biology, physics and engineering and expect them to survive contact with data. We want answers we can measure, not answers we admire. In your first 90 days, you will own one well-defined research problem: define an objective and credible baselines, run a reproducible experiment and find where the method fails.

Who we are

Nature keeps records without writing anything down. Tree rings hold the droughts of past centuries, a salmon finds the river where it was born, and a forest remembers a fire in what grows back. Soon, machines will work all around us: in our homes, on our streets and out in nature. What will make them more useful, and better at working with people? Perhaps the ability to remember.

Superstrata is building fundamental, nature-inspired innovations in memory that unlock up to twice the output from the same GPU. We're already talking to some of the world's largest players in AI infrastructure and physical AI. You'll be one of the first ten employees, working directly with the CEO: an avid kitesurfer who trained in lithography in the early days of TSMC's fab labs. We're an unlikely team of good humans, working to build brilliant technology that serves life on Earth. First cheques are in; we aim to raise a big round by EOY.

What you'll be doing

• Turn open research questions into precise, testable problems. Make the assumptions explicit.

• Design experiments with credible baselines and clear stopping rules.

• Connect mathematical reasoning to measured results, and check that derivations match their implementations.

• Report negative results and reject gains that do not hold up.

What we need to see

• Strong applied mathematics, linear algebra, optimisation or experimental design.

• The ability to implement an experiment in Python and explain the result.

• Comfort moving between a derivation, code and measured behaviour.

• Willingness to find the flaw in an elegant argument, including your own.

• A degree in mathematics, physics, computer science, statistics or a related field, or equivalent research experience.

Ways to stand out

• Information theory or statistics.

• Experimental design or decision-making under uncertainty.

• Applied physics or efficient computing.

• Published research, a preprint or open-source code. A PhD may be relevant; it is not the only way to show research ability.

Signs it could be a great fit

• You decide what would count as failure before you run the experiment.

• A clean negative result interests you as much as a positive one.

• You want to work with people who think differently from you, and to let the evidence decide.

Signs it may not be the right fit

• You want a fully specified research agenda before you start. Much of our work begins as an open question.

• You want to manage people from day one. Everyone on the technical team builds.

• You prefer theory that never meets a measurement.

Our interview process

We move fast. If selected for an interview, you will have a first conversation within days, and we will tell you where you stand after every step.

• Application review and initial screening.

• Initial conversation (30 minutes).

• Technical conversation about a paper, notebook or project you did (60 to 90 minutes).

• Paid work trial on a bounded problem.

Practical details

• Location: London, United Kingdom.

• Working pattern: Hybrid. We work together in person in London regularly.

• Engagement: Full-time.

• Reporting: To the CEO.

• Salary: Your base salary depends on your experience. The range is £60,000–£100,000 for this role as specified.

• Equity and benefits: Equity in Superstrata: a share in what you help build. 25 days' paid holiday plus UK bank holidays, and a workplace pension.

• Compute: You will have access to plenty of compute.

• Right to work: You must already have the right to work in the UK. We cannot sponsor a visa at present. If you will need sponsorship in future, tell us when you apply.

• How to apply: Email careers@superstrata.ai with a short introduction and one paper, notebook or technical project: a link to it, or a PDF. Include your contribution, a limitation or negative result, and what it taught you. Please do not share a previous employer's confidential material.

• Closing date: Rolling. We review applications as they arrive and expect to hire quickly.

• Use of AI in recruitment: Superstrata uses AI tools to help organise applications. A person makes every hiring decision. Our candidate privacy notice is at superstrata.ai/privacy/candidates.

Superstrata is an equal opportunity employer. If you need an adjustment at any stage of the process, tell us.

Pay: £60,000.00-£100,000.00 per year

Benefits:

• Company events
• Company pension
• Employee stock ownership plan
• Employee stock purchase plan
• Work from home

Work Location: Hybrid remote in England