Industry-PostDoc: Machine Learning for Material Science (Closed)
Deadline: 31st January 2024 (cannot guarantee applications to be considered afterwards)
You will be working closely together with scientists in the materials space to build OpenSource machine learning models (OptApps) that will inform laboratory experimentation schedules, anywhere from complete-manual to fully self-driving.
Specifically, there will be an emphasis to identify, implement and test models for concrete sciences, e.g. as described in https://arxiv.org/abs/2310.18288
Try our 3 minute tutorial to see what that looks like: https://matterhorn.studio
Check out this 15 minute talk to learn more about the OptStore and its OptApps:
You work will focus on:
- Researching the right Bayesian Optimisation techniques for a variety of experimentation challenges: multi-fidelity, multi-source, multi-step (generally known as ‘grey-box’ methods, see https://arxiv.org/abs/2201.00272)
- Implementing these models as OpenSource "OptApps" on the OptStore, to be shared with the material science community, focusing on ease of usability and interpretation
- Validating the effectiveness of the models and tackling deeper research challenges, with the opportunity to publish.
You must have relevant experience in Machine Learning, specifically Bayesian Optimisation, at MSc/PhD level to complete the above work.
Location: Hybrid/Oxford/London (anywhere in the UK)
Length: In a first instance, limited to 8 months, from 1st of April to 30th of November, with an option for extension or permanent position.
Matterhorn Studio is leading a paradigm shift towards peer-reviewed plug-in Bayesian Optimisation.
We’re looking forward to see how we can shape the future of material science together with you.
Contact Jakob at jakob@matterhorn.studio with a CV and a few ideas of who you’d approach the above work.
Get in touch
We will happily guide you through the emerging space of data-driven material discovery. We look forward to learn from your experience and problem space.
