Digital Twin & Surrogate Modelling Intern M/F
Confirmed on the employer's own hiring board on Nov 2, 2026. First seen by Alion on Oct 19, 2026. Alice & Bob scores A on the Alion truth index.
PURPOSE OF THE ROLE
The intern will investigate whether high-fidelity simulations, fast surrogate models and experimental data can be combined into a digital twin of Alice & Bob's cat-qubit devices. As these devices scale up, the predictive accuracy of physical and numerical models becomes harder to guarantee, and measurements from chips in our cryostats can validate and refine the models.
Data assimilation is the general framework that combines simulations and experimental data: it updates model states from measurements, accounting for model and measurement uncertainties, so the models stay accurate as new observations arrive. High-fidelity simulations, however, are costly, which limits how many the team can run. Surrogate and multi-fidelity models ease this limit by delivering high-fidelity results at a fraction of the cost, and they integrate into data assimilation. Faster and more accurate simulations then let the team explore more scenarios within the same compute budget.
The intern will build a proof of concept on a representative use case, defined jointly with the team during the first weeks. The deliverables are a validated prototype and the methodological and technical building blocks of a future digital-twin framework. Reporting to the Scientific Computational Engineer, the intern will work with the Simulation Engine team and with the teams producing experimental data.
This internship will start in February 2027 for a duration of 6 months!
Responsibilities:
- Develop and apply surrogate models - Build fast approximations of high-fidelity simulations - Assess their accuracy, uncertainty and domain of validity
- Combine models and experimental data - Implement data assimilation methods linking measurements to simulations - Quantify model and measurement uncertainties
- Run and analyse simulations - Use the team's high-fidelity simulation tools to generate and interpret results - Compare simulation outputs with experimental data
- Develop scientific software - Write clean, tested and documented code (Python / JAX) within the team's codebase - Contribute to the design of reusable components for a digital-twin framework
- Communicate and collaborate - Present results and methodological choices to the team - Work with experimental and design teams to understand data and needs
Requirements:
- EDUCATION: Final year of a Master's degree (M2) or engineering school in physics, applied mathematics or scientific computing
- LANGUAGES: Proficient in English
- EXPERIENCE: No professional experience required, since coursework and research projects can supply the skills above. A previous internship or research project in scientific computing or numerical simulation is appreciated.
- Background in engineering
- Solid foundations in numerical analysis and numerical methods
- Knowledge of data assimilation methods, including uncertainty quantification
- Knowledge of reduced-order modelling
- Experience with open-source scientific simulation software
- Proficiency in Python, with strong experience in JAX
- Good software engineering practices, namely version control (Git), testing, documentation, CI/CD, and verification & validation (V&V)
- Rigour, autonomy and curiosity, and the ability to explain technical choices clearly
A candidate with an interest in scientific software who can structure an open-ended problem and make progress autonomously while keeping the team informed.
Nice to have:
- Statistics or Bayesian inference
- Experience with PyTorch or TensorFlow
- Deep learning
- Quantum physics
- Circuit engineering
- Rigorous use of agentic workflows for scientific research and implementation