Master Thesis – Data-Efficient Exploration

The Materials Modelling team accelerates the development of next-generation cutting tool materials through advanced modelling, simulation, data analytics, and AI. We work across multiple length scales, from atomic-scale phenomena to industrial processes, transforming scientific understanding into practical solutions that support materials design, process development, digitalization, and sustainable innovation. By combining materials science, computational methods, and data-driven approaches, we help reduce development time, improve decision-making, and enable virtual product development.

We’re now looking for a student who wants to complete their master's thesis with us, focusing on data-efficient exploration of compositions and mechanical properties using first-principles calculations and active learning.

Background and purpose

The computational design of materials for cutting tools requires predictive models that can describe mechanical-property trends across multicomponent composition spaces. Existing first-principles databases provide valuable information on elastic constants and related mechanical-property descriptors, but their coverage is generally non-uniform. In particular, predictions in sparsely sampled regions may be associated with substantial uncertainty.

Work description

This thesis will develop active-learning strategies to reduce prediction uncertainty by identifying a limited but sufficient set of informative compositions for additional autoQMAS calculations. The newly generated data will be iteratively incorporated into a database to update and improve the predictive models.

The objective is to achieve a predefined confidence level across the compositional region of interest. At the same time, the number of additional calculations should be minimized and compositions with the best trade-off between hardness and toughness identified.

Project plan:

Weeks 1-2: Literature review and project definition

Weeks 3-4: Existing workflow and database - baseline model and uncertainty assessment

Weeks 5 -12: Active-learning method development and implementation

Weeks 13-15: Final analysis and documentation

Weeks 16-20: Writing thesis and preparing presentation

Location

The position is based in Västberga, Stockholm.

Profile

We are looking for a Master's student with a strong foundation in physics, materials science, applied mathematics, or computational engineering. The ideal candidate has experience with Python programming, scientific computing, data analysis, and machine learning. Knowledge of atomistic simulations, autoQMAS calculations, density functional theory (DFT), or computational materials science is meritorious. The project requires curiosity, initiative, and an interest in combining physics-based modelling with AI-driven approaches for accelerated materials design.

Our Culture

At Sandvik, we work with advanced technology and exciting innovations, but it is our people who are the true key to our success. We believe that diversity creates a better environment and that inclusion is essential to achieving strong results. This means supporting one another, sharing knowledge, and embracing each other's differences.

To learn more about us, we encourage you to visit our website, LinkedIn, or Facebook.

Scope

The master’s thesis project corresponds to 30 ECTS credits and is expected to be carried out full-time over approximately 20 weeks. The project is planned to start in January 2027.

Contact Information

For more information about the position, please contact Aayush Sharma, Senior R&D Professional +46 (0)70 616 79 47

We have carefully selected the recruitment channels and marketing methods we wish to use and kindly, but firmly, decline any additional contacts regarding these matters.

Application

Please submit your application no later than December 1, 2026. Click Apply and attach your CV and cover letter. Please note that we do not accept applications via email. Job ID: R0097769.