Software Engineering, Data Science
Posted on Sunday, August 27, 2023
Stellarator optimization is a rapidly moving field and is being revolutionized by advances in computation and machine learning. New techniques have unlocked incredible potential to discover device geometries that achieve transformative levels of performance.
Machine Learning holds significant promise for advancing stellarator research and design. In particular, ML-based surrogate models offer a cost-effective approach to quickly explore the design space, overcoming challenges posed by high-fidelity, noisy physics simulations. Moreover, ML is starting to play a critical role in bridging gaps within fusion physics: it enables the development of data-driven reduced models for key phenomena and quantities currently lacking comprehensive theoretical frameworks.
Proxima Fusion’s StarFinder stellarator optimization framework, developed to design stellarators with a complex set of both physics and engineering constraints, paves the way for the application of AI to the field. Proxima Fusion’s database of stellarator configurations is reaching one million entries, offering unique opportunities for model training.
As an ML engineer, you will support the team’s effort to apply cutting-edge machine learning models to our stellarator optimization framework, accelerating the search of simpler and more robust stellarators that achieve economic viability. You will get to work on technically challenging problems and deploy your skills at the forefront of fusion R&D bringing us closer to fusion power plants.
We are looking for candidates with backgrounds including:
Technical ML proficiency: you possess extensive familiarity and hands-on experience with modern machine learning software frameworks (such as JAX and pytorch).
Software engineering expertise: you have demonstrated the ability to design sophisticated software systems and quickly comprehend complex trade-offs in system design.
Holistic grasp: you can display a comprehensive understanding of the realm of scientific machine learning, coupled with a keen interest in curating datasets and training models tailored for scientific applications.
As an ML engineer, you will:
Curate datasets: Establish and manage pipelines for generating and refining datasets that underpin our scientific AI/ML endeavors.
Develop end-to-end pipelines: Develop streamlined pipelines for automated model training, evaluation, and deployment within a production environment.
Design state-of-the-art tools: Devise and implement cutting-edge tools and technologies to facilitate fusion research and accelerate iteration, spanning from data collection to model optimization.
Improve training data: Explore novel techniques for data augmentation, data filtering, and data generation to enhance the quality of training data.
Bring models to life: Collaborate closely with fusion scientists and software engineers to seamlessly integrate models into our optimization stack.