Member of Technical Staff, Applied AI
Member of Technical Staff, Applied AI
> ROLE 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 engineers to apply AI, ML, and simulation to real-world customer physics and engineering problems. 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.
> CORE RESPONSIBILITIES
- Work with customers to apply AI and ML to hard, real-world physics problems. Build ML models, simulations, and digital twins that beat traditional engineering workflows on accuracy, speed, or both, pair them with agentic optimization loops, and ship results customers can actually run in production.
- Build the demos and engagement-specific tooling that make PSI's physics-AI capabilities tangible. Working artifacts and live dashboards, not slide decks. Every engagement ends with something a customer can run and see results from.
- Translate real-world data into clean inputs for ML and simulation pipelines: sensor telemetry, design documents, operational logs, public data feeds. Define schemas, build ingestion pipelines, make every engagement's data usable in days, not months.
> HARD REQUIREMENTS & SPECS
- Five or more years building ML systems in production with grounding in applied physics, computational science, or engineering, at companies or labs known for scientific rigor. You have written ML code that solved a real physics or engineering problem, not just synthetic benchmarks.
- Strong physics literacy in at least one quantitative domain. You can read a domain paper, hold a technical conversation with a senior domain engineer, and reason about non-linear trade-offs between accuracy, speed, and extrapolation risk.
- Demonstrated ability to build ML models, simulations, or digital twins for physics or engineering problems. You have shipped systems that held up under real-world distribution shift, not just on the training set.
Is the AI extraction inaccurate? Report an issue