Internship: Machines that design themselves; inventing green nanomechanical systems using large language models

Large-language models (LLMs) have impressive capabilities, such as automatically generating code, writing poetry, or summarizing text; but can they be used to automate the design of mechanical machines? In the hypersmart matter group at AMOLF, we want to answer this question in the context of smart nanomechanical systems.

Nanomechanical systems lose very little energy when they vibrate, a property that we want to use to make greener, more user-friendly gadgets. Examples of such would be wearables like smartwatches that do not require batteries, or implantable medical sensors that can operate permanently without being recharged. However, finding out how to combine mechanical elements (masses, springs, dampers, nonlinear interactions) to produce the desired advanced functions (i.e., to build a pacemaker or a wearable fitness monitor) is extremely hard – due to the very large design space (there are infinitely many systems one could build!) and due to the complex relation between device structure and performance. To solve this problem, we want to take advantage of large language models to automate the search, testing, and benchmarking of possible designs.

The goal of this internship is to build a software loop integrating an open-source LLM, together with a mechanical simulation platform, so that the large language model generates candidate mechanical systems, simulates them, and evaluates their performance (using e.g., automated code generation, train-of-thought prompting, etc.). We will use our GPU cluster to fine-tune the LLM using reinforcement learning, so the system continuously learns and improves through the experimentation process.

The internship is open to master’s degree candidates from a range of backgrounds, including Engineering, Physics, Computer Science, Mathematics or a related field. We are looking for ambitious candidates who are excited about building something that has never been done before. Specific experience with programming languages, software or equipment is not mandatory, but motivation and desire to learn is.
The internship must be a mandatory part of your curriculum. You have a nationality of an EU-member state and/or you are a student at a Netherlands University. You must be available for at least 6 months.

Dr. M. Serra Garcia
Group leader Hypersmart Matter
E-mail: [email protected]
Phone: +31 (0)20-754 7201

You can respond to this vacancy online via the button below.
Please annex your:
–  Resume;
–  List of followed courses plus grades.

Applications will be evaluated on a rolling basis and as soon as an excellent match is made, the position will be filled.

Online screening may be part of the selection.

AMOLF is highly committed to an inclusive and diverse work environment. Hence, we greatly encourage candidates from any personal background and perspective to apply.

Commercial activities in response to this ad are not appreciated.