Overview

Pacific Fusion is seeking an exceptional research intern to join our team and create the path to commercial fusion energy.

At Pacific Fusion, the machine learning team works across several computational teams to accelerate and improve our critical computational capabilities. In this role, you will contribute to the development of machine learning systems that make our multi-physics simulators faster, higher fidelity, and overall more impactful to their internal users. You will collaborate with computational physicists, target designers, and experimental physicists to find the highest leverage research work that enables us to move faster and more efficiently.

We are looking for candidates interested in a 16 to 24 week, paid, in-person internship based in our Fremont, CA office.

Responsibilities

The main responsibility of a research intern in this role is to carry out research towards developing novel ML methods for accelerating expensive CFD-like simulations used for fusion target design. You will be expected to (1) read through ML and physics literature that is relevant to your project, (2) design, implement and train novel ML architectures, and (3) produce internal and/or external reports of your work in the form of whitepapers or publications to top-tier research venues.

Strong candidates will have the following core competencies:

1. ML and physics expertise: research interns in this role work at the intersection of machine learning and computational physics, and must be able to reason across both domains (although it is fine to have deeper background in one or the other).
2. Communication: you’ll be working at the intersection of multiple highly technical disciplines, and you will collaborate with a diverse set of subject matter experts across computational, theoretical, and experimental plasma physics. You will be expected to work with your colleagues to translate these problems into machine learning ones.
3. Agency: we have a lot of exciting machine learning problems at Pacific Fusion. You’ll work in an iterative, kind, and deeply collaborative environment to new ML capabilities in computational plasma physics. You’ll be given the freedom as well as the responsibility to push directions that you believe in and that have the chance to provide the greatest internal and external impact.

Qualifications

1. Currently in the process of obtaining a PhD in Machine Learning, computational physics, plasma physics, or related fields.
2. Hands-on and mature experience in deep learning algorithms, techniques, and frameworks such as PyTorch and JAX.
3. Software development knowledge and experience in Python and the relevant scientific and computational stack (numpy, scipy, matplotlib, etc).
4. Experience on applying machine learning to model interesting physical problems.
5. An interest in pushing the state of the art on problems at the intersection of physics and machine learning, including building surrogate models of physical systems, learning and optimizing PDEs, unsupervised learning for physical data, automatic differentiation, bayesian inference, and so on.
6. Proven track record of publications at ML venues such as NeurIPS, ICML, ICLR, TMLR, and/or in top-tier physics journals.