Photo: Shakil Ahmed / Pexels
Most machine learning jobs ask you to predict what a shopper will click next, or which ad someone might open. This one is different. The University of Exeter is hiring a researcher to predict the ocean itself — its temperature, its currents, where the next measurement should be taken, and how much you can actually trust the forecast.
The post is a Postdoctoral Research Associate / Fellow in Machine Learning Ocean Forecasting and Uncertainty Quantification, based at the university's Penryn Campus in Cornwall. It is a real, paid, full-time job — not a fellowship award, not a studentship — funded by UK Defence Innovation and run with two partners who know the sea from opposite ends: the UK Met Office, and Oshen, a company that builds autonomous ships to take ocean measurements. The pay starts at £35,302 a year, or £44,352 if you come in at the senior Fellow level. Applications close at 23:59 on 10 December 2026.
If you hold a PhD in machine learning, statistics, oceanography, mathematics or engineering — or you are close to finishing one — and you have ever wanted your models to do something that matters outside a dashboard, this is worth a proper look.
At a glance
| Role | Postdoctoral Research Associate / Fellow in Machine Learning Ocean Forecasting and Uncertainty Quantification |
| Employer | University of Exeter — Faculty of Environment, Science and Economy, Department of Engineering |
| Location | Penryn Campus, Cornwall, United Kingdom (hybrid working) |
| Contract | Full-time, fixed-term, 12 months: 1 January 2027 – 31 December 2027 |
| Salary | £35,302 per year (Grade E, Associate) or £44,352 per year (Grade F, Fellow) — starting salaries, plus staff benefits |
| Advertised | 30 October 2026 |
| Deadline | 10 December 2026, 23:59 (UK time) |
| Job reference | 1817 |
| Application fee | None |
What is this job actually about?
Weather forecasting got very good, very fast, once machine learning entered the picture. Ocean forecasting is harder. The ocean is observed far less densely than the atmosphere — you cannot put a weather station every few kilometres across the Atlantic — so the models have to do more with less data, and say honestly how uncertain they are.
That is the heart of this project. The team is building a new forecasting system that learns from real ocean observations rather than from physics simulations alone. Your job would be to help build it, test it, and prove it works. Three pieces of the work stand out.
First, uncertainty quantification. A forecast without an error bar is a guess dressed up as an answer. The project is building uncertainty propagation into a three-dimensional machine learning pipeline, so the system can say how reliable each prediction is — and where it is guessing. If you have worked with Bayesian methods, ensembles, or calibration, this is your territory.
Second, adaptive sampling. When your measurements come from autonomous vessels that can only be in one place at a time, where you send them matters. The role includes designing algorithms that decide where the next measurement would add the most information — the ocean, in effect, telling the ships where to go.
Third, validation against the real world. Models get tested on synthetic ocean datasets and on real data — MEOP-CTD datasets and measurements collected by autonomous vessel deployments — alongside researchers at the Met Office. The work ends in published papers and conference talks, not in a model that sits on a shelf.
Day to day, expect Python, simulation code, staged verification and validation of the forecasting framework, and a lot of time spent with observational data. You will work with Dr Ajit Pillai and Dr Ian Ashton at Exeter, plus Met Office research teams.
Who can apply?
Exeter is hiring at two levels, and which one fits you depends on your experience. Read both before you decide — applying at the wrong level is the easiest avoidable mistake here.
Postdoctoral Research Associate — Grade E (£35,302)
You need a relevant PhD, or to be nearing completion of one, in oceanography, machine learning, statistics, mathematics, engineering or a closely related field. Equivalent qualifications and professional experience can also count. Beyond the degree, the university wants to see that you know your discipline and its research methods, that you can communicate progress and results clearly in writing and in person, and that you can help write proposals for external funding. This is the right level if you are finishing your PhD now, or you have recently finished and your publication record is still growing.
Postdoctoral Research Fellow — Grade F (£44,352)
The Fellow level asks for more. You need a relevant PhD plus real specialist expertise: advanced work in uncertainty quantification and propagation, experience validating complex computational models against real observational data, and a record of building open-source codebases or reproducible scientific software. You should also be able to develop research objectives and proposals, work across teams, supervise other researchers, and lead when needed. If that reads like your CV, apply at this level — the £9,050 difference in starting salary reflects a genuine step up in responsibility.
One honest note: the advert says nothing about visa sponsorship. If you would need a visa to work in the UK, do not assume anything — email the HR Helpdesk before you spend hours on the application, and ask directly.
What does it pay?
Let's be straight about the money, because it matters. The starting salary is £35,302 a year at Associate level and £44,352 a year at Fellow level, with staff benefits on top. Your final grade and salary depend on your qualifications and experience.
Compared with an industry machine learning job, that is modest — a senior ML engineer in London or the US can earn three or four times as much. But this is a university research post, and the trade is different: you publish, you work with the Met Office and a defence-funded programme, you build a public record in a field — uncertainty-aware ocean forecasting — where very few people have deep experience. That record is worth real money later, whether you stay in research or move into industry climate and forecasting roles. And Penryn is a small town in Cornwall, where your rent goes a lot further than it would in London.
The contract is twelve months, 1 January to 31 December 2027. That is the other honest trade-off. A year is enough to do serious work and publish from it, but it is not long-term security, and you should go in with your eyes open about what comes next.
How to apply
Applications go through the University of Exeter's own careers portal — not by email, and not through a jobs board. Here is the process:
- Read the full job description and person specification on the Exeter careers portal before you write anything. The advert says this plainly, and it is good advice — the person specification is what the shortlisting panel actually scores you against.
- Decide your level. Grade E (Associate) or Grade F (Fellow)? Pick the one your experience genuinely matches, and shape your application to its requirements.
- Prepare your application — CV, and evidence that maps to the requirements. Highlight machine learning work, uncertainty quantification, Python, computational modelling, oceanographic data analysis and scientific software development where you have them. For the Fellow level, make your open-source and supervision record easy to find.
- Submit online before 23:59 on 10 December 2026, quoting job reference 1817. Exeter's adverts close at 23:59 on the stated date — do not leave it to the last evening, because the portal will not wait for you.
- Ask questions early. Role questions go to Dr Ajit Pillai (a.pillai@exeter.ac.uk). If the portal gives you trouble, the HR Helpdesk (hrhelpdesk@exeter.ac.uk) is the contact.
There is no application fee. Nobody will ask you to pay anything to apply — if anyone does, it is not the university.
Important dates
| Advertised | 30 October 2026 |
| Application deadline | 10 December 2026, 23:59 (UK time) |
| Job starts | 1 January 2027 |
| Contract ends | 31 December 2027 |
The deadline gives you just over five weeks from the day this article was published — enough time to put together a careful application, but not enough to drift. If this role fits you, start this week.
Official links
Apply on Official Website — University of Exeter Careers portal. Search for job reference 1817 (Postdoctoral Research Associate/Fellow in Machine Learning Ocean Forecasting and Uncertainty Quantification) to open the vacancy and apply.
More from the employer: University of Exeter — official website.
Should you apply?
Apply if you are finishing — or have recently finished — a PhD in machine learning, statistics, oceanography or a related field, you write solid Python, and the idea of putting error bars on ocean forecasts genuinely interests you. Apply at Fellow level if you already have the uncertainty-quantification expertise and the open-source record to back it up.
Skip it if you do not have a PhD or near-complete doctorate — the advert is firm on that. Skip it if you need a multi-year contract right now, or if you cannot be based around Penryn in Cornwall with hybrid working. And skip it if what you really want is the highest salary your ML skills can command — industry will pay more, and this role's reward is the research record, the Met Office collaboration, and the niche expertise, not the pay cheque.
Still earlier in your journey? If you are a student building towards a research career, our Ai2 Research Internship 2027 in Seattle is a paid way to get research experience, and the GovAI Summer Fellowship 2027 in London is another route into serious AI research. And if you need income while you study, these 9 AI side hustles for students are a practical starting point.
Frequently asked questions
Do I need a PhD to apply?
Yes, or to be nearing completion of one, for the Associate level. A relevant PhD is required for the Fellow level. Equivalent qualifications and professional experience may also be considered, but a doctorate — finished or nearly finished — is the expected route in.
Is there an application fee?
No. There is no fee to apply for this job, and the University of Exeter does not charge candidates at any stage of its recruitment.
Is visa sponsorship available?
The advert does not mention visa sponsorship. If you need sponsorship to work in the UK, contact the university's HR Helpdesk (hrhelpdesk@exeter.ac.uk) before applying to confirm your position.
Can I apply for both the Associate and Fellow levels?
The university advertises one vacancy filled at either level, depending on the successful candidate. Read the person specification for both, decide which level your experience matches, and apply at that level.
Where exactly is the job?
Penryn Campus in Cornwall, in the south-west of England, with hybrid working — part of the week on campus, part from home. It is not a London role, and the salary should be judged against Cornwall living costs, not London ones.
What happens after the deadline?
Applications close at 23:59 on 10 December 2026. The post starts on 1 January 2027, so shortlisting and interviews will move quickly after the closing date.
Ocean forecasting is one of those fields where machine learning is not a nice extra — it is the only practical way to fill the gaps between sparse measurements and the forecasts people depend on. If your PhD gave you the tools for that problem, this is a year that could shape the rest of your research career. The deadline is 10 December 2026. Apply on the official Exeter careers portal, reference 1817.
All details in this article were checked against the University of Exeter's vacancy advert (job reference 1817) on 2 November 2026. Deadlines and details can change — always confirm on the official careers page before applying.