Machine Learning Scientist: Foundation Models

Contract Type: Fixed Term Appointment – until March 2027

Interview Date: W/C 25th November 2024

Purpose:

Sea ice is a key global climate indicator for describing the changing climate, it influences local climate conditions and provides important habitat and breeding grounds for keystone species in the polar regions. Passive microwave (PM) derived sea ice concentration (SIC) datasets are arguably the most widely used datasets for monitoring trends in sea ice, providing an unrivalled 50-year observational record sea ice conditions. The dataset is widely used by scientists, local communities, politician, policy makers and industry in the polar regions. Despite its fundamental importance, it is widely acknowledged within the sea ice community that the present form of PM-derived SIC products are imperfect, due to a lack of rigour in their assessment of uncertainty. The British Antarctic Survey (BAS) is looking to employ a machine learning scientist to join our programme focused on quantifying the spatial and temporal uncertainties in the Antarctic PM-derived SIC record. Working in a team of machine learning researchers, software engineers, remote sensing specialists, and sea ice experts, the successful candidate will lead the development of a groundbreaking self-supervised foundation model to automatically generate high resolution SIC (SIC-HR) products from the large archive of multispectral and synthetic aperture radar satellite imagery. Through collaboration with project partners in the US National Snow and Ice Data Centre, Norwegian Ice Service, Alan Turing Institute, and others, these SIC-HR products will be used to directly quantify the uncertainty in the PM-derived SIC record.

The successful candidate will join the Artificial Intelligence team at BAS, which includes research software engineers and machine learning researchers with a passion for applying AI techniques to answer some of the most pressing questions surrounding the earth’s cryosphere. BAS is an interdisciplinary organisation, and the post-holder will have opportunities to develop strong collaborations with remote sensing experts, data scientists, and sea ice physicists across BAS and beyond. Support and training will be provided to enable the successful candidate to make use of the opportunities BAS has to offer and to develop their scientific career. This post is aligned with a strategic initiative to deliver on the new BAS Science Strategy.

The start date for this role is flexible, but ideally spring 2025. If this role appeals to you, please get in touch if you have any questions, whatever your background. Informal enquiries about this post are very welcome and should be addressed to Martin Rogers (marrog@bas.ac.uk).

 Duties:

• To train and deploy a self-supervised foundation model of the polar regions to automatically generate high resolution SIC (SIC-HR) products from satellite imagery.
• Collaborate with national and international colleagues to quantify the uncertainty in the passive microwave sea ice concentration record using these SIC-HR products.
• Explore potential future collaborations for the wider application of the polar foundation model developed.
• To champion reproducible science and open-source infrastructure to empower the global environmental research community
• To represent BAS to key stakeholders, such as funding agencies and the European Space Agency.
• Disseminate research findings through high-impact peer reviewed journal articles, reports, and presentations at national and international conferences, clearly highlighting novel results and their impact.
• To work both independently and with small teams to contribute ideas, direct projects, and mentor other scientists.
• To engage with the academic life and activities of British Antarctic Survey, NERC and UKRI.

Skills Specification

Qualifications

Essential:

• A PhD in a relevant field, plus post-doctoral experience. Equivalent relevant experience will be considered.

Skills & Experience

Essential:

• Expertise in developing and training state-of-the-art machine learning tools.
• Proficiency in one or more modern statistical programming languages used in research in data science and machine learning, such as Python.
• Experience of working with remote sensing data, satellite imagery or other large environmental datasets and writing scalable code.
• Excellent track record of disseminating research results through the publication of high-quality reports, peer-reviewed literature, and conference presentations.
• Previous experience of managing oneself and resources/components of projects to meet objectives and deadlines.
• Proficient in written and spoken English language. In particular, the ability to present science in a clear and compelling way to a variety of stakeholders.

Desirable:

An interest in sea ice trends and dynamics.

Contribution and engagement with open-source software communities, including experience in collaborating via version control.

The role holder will be required to have the appropriate level of security screening/vetting required for the role. UKRI reserves the right to run or re-run security clearance as required during the course of employment.

Who are we?

British Antarctic Survey (BAS) delivers and enables world-leading interdisciplinary research in the Polar Regions. Its skilled science and support staff based in Cambridge, Antarctica and the Arctic, work together to deliver research that uses the Polar Regions to advance our understanding of Earth as a sustainable planet. Through its extensive logistic capability, BAS facilitates access for the British and international science community to the UK polar research operation. Numerous national and international collaborations, combined with an excellent infrastructure help sustain a world leading position for the UK in Antarctic affairs. British Antarctic Survey is a component of the Natural Environment Research Council (NERC). NERC is part of UK Research and Innovation www.ukri.org.

We employ experts from many different professions to carry out our Science as well as to keep the lights on, feed the research and support teams and keep everyone safe! We aim to attract the best people for our jobs. If you are looking for an opportunity to work with amazing people in amazing places, then British Antarctic Survey could be for you.

Our vision

At BAS, our vision is to be a world-leading centre for polar science and operations. Making our vision a reality depends on the excellence and diversity of our staff. We are committed to creating a workplace where all our staff can flourish and succeed. BAS is a Disability Confident employer, we are proud to hold a silver Athena Swan award and we are a member of enei, the Employers Network for Equality & Inclusion.

Working at BAS

Choosing to come to work at BAS means that you will have access to a whole host of benefits from a defined benefit pension scheme, excellent holiday entitlement, access to employee shopping/travel discounts and salary sacrifice cycle to work scheme. You can find out more about our benefits here.

We appreciate the importance of achieving work-life balance and support this with several family and carer-friendly policies. Plus, a flexible working policy for those who may wish to amend their working pattern or arrangement. We also operate a hybrid working policy to support the balance of working time between the office and home.