Member of Technical Staff, Applied AI

Overview

Physical Superintelligence is a startup with roots at Google, NVIDIA, Harvard, Meta, MIT, Princeton, Oxford, Johns Hopkins, Cambridge, and the Perimeter Institute building AI systems to discover new physics at scale. We are seeking applied physicists, engineers from the traditional disciplines (mechanical, electrical, aerospace, chemical), and computational scientists to put AI to work on real engineering systems: thermal and fluid systems, electrical power distribution, structures, and the coupled models that tie them together. Models that beat the solvers they replace, and AI that designs rather than only evaluates. Applied AI at PSI is applied physics plus AI.

Our mission is to discover and commercialize transformative physics breakthroughs at scale with artificial superintelligence, safely, verifiably, and for broad public benefit.

The last century's golden age of physics gave us transistors, lasers, and nuclear energy. We believe artificial superintelligence will unlock the next one. We're creating the infrastructure to industrialize scientific discovery and usher in this new era.

We have one product: new physics, at scale.

Role and Responsibilities

• Model real physical systems with AI. Build simulations, surrogates, and coupled multiphysics models of engineering systems, thermal, fluid, electrical, and structural, that beat traditional workflows on accuracy, speed, or both, and validate them against solvers and measured data.

• Model each domain at engineering depth. Conjugate heat transfer and airflow, grid interconnection and protection studies of power distribution and the dynamics of the power electronics behind it, structural and vibration analysis, electrochemistry. Pair the classical studies with learned models and optimization.

• Learn from simulators. Train models over classical simulation that run orders of magnitude faster than the solvers they replace and still hold up on inputs outside the training data. Training-set accuracy is not the bar.

• Close the loop from analysis to design. Use models and agents to search a design space, not only to score a design someone else proposed.

• Make simulation legible to agents. Decide where solver fidelity is required, where a learned model is enough, and how the two work together.

• Ship to customers. Each engagement solves a customer's actual problem, ends with something they can run, and leaves behind a capability we reuse on the next one. Publish the methods where it serves the mission.

What We're Looking For

• A PhD, or an equivalent research record, in mechanical, electrical, aerospace, or chemical engineering, applied physics, computational science, or machine learning for science. You have modeled a physical system that mattered and validated the model against ground truth: a surrogate against a solver, a simulation against a facility, a model against measured data. You can say how you knew it was right and what you did when it was not.

• Depth in at least one engineering domain. Thermal and fluid (CFD, conjugate heat transfer), power systems (grid interconnection studies for large loads including ride-through, protection, and the dynamics of UPS, BESS, and converters), structural (FEA), or electrochemistry, with the tools of that domain (OpenFOAM, SU2, ETAP, PSS/E, PSCAD, OpenDSS, pandapower, CalculiX, Code_Aster, or comparable) and the standards it works to. You know where the models lie.

• Hands-on computational engineering. You have run and extended CFD, FEA, power system, or multiphysics codes, you understand discretization, convergence, and boundary conditions, and you know the trade-offs between a high-fidelity solver and a faster learned alternative.

• Machine learning you have applied yourself: PyTorch or JAX, training runs you designed and debugged, models that held up outside the training set. You do not need a top-venue publication record. You do need to have shipped ML on a physics problem rather than on a synthetic benchmark.

• You have put a model into a system other people depend on: behind an API, versioned, reproducible, with tests that catch a wrong number and not just a crash. You are comfortable in HPC and cloud GPU environments.

• Judgment on real-world problems. You can scope a problem with a senior domain engineer, tell them what is credible and what is not, and deliver working artifacts rather than slide decks.

• Nobody is expected to be expert in all of the above. We hire for strengths, not for the absence of weaknesses: strong in most of it, deep in at least one domain. You start on an open-ended problem without waiting to be told how, pick up an unfamiliar domain when the work needs it, are comfortable with ambiguity, and work well across disciplines and as part of a team.

Nice to Have

• Neural operators, physics-informed networks, differentiable simulation, or uncertainty quantification for learned models.

• Experience with open-source simulation codes (OpenFOAM, OpenDSS, pandapower, CalculiX, or comparable) rather than only commercial packages.

• Domain work in thermal and fluid systems, power systems and power electronics, structures, or electrochemistry.

• Optimization over design spaces: mixed-integer, derivative-free, or Bayesian search.

• Peer-reviewed publications in computational physics, engineering, or ML for science.

How We Work

We hold a high technical bar and give people full ownership of their work, from spec to ship to on-call. We write contracts before logic, test against real systems instead of mocks, and favor simple designs that ship over clever ones that do not. Our development process is AI-native: we work with agentic coding tools daily, write specs that are legible to humans and agents alike, and lead with leverage.

Location and Compensation

This role is based in Boston. We will consider remote candidates on a case-by-case basis. We offer competitive compensation including salary, benefits, and meaningful early-stage equity. We evaluate on physics depth, modeling judgment, ML fluency, and shipping velocity. We are an equal opportunity employer and value diverse perspectives in building platforms for AI-driven discovery.