Who we are
SIMLab works in close collaboration with academic and industrial partners, with a strong emphasis on practical applications and ensuring the real-world impact of our research.
Core Areas
Differential Programming
We heavily rely on general differentiable programs, enabling the integration of domain knowledge and complex structures into learning models.
Advanced Data Processing
Handling large-scale and complex datasets using state-of-the-art technologies, scaling pipelines to very large-scale datasets.
Hybrid Modeling
Combining data-driven approaches with traditional simulations using differential programming methods, leveraging the strengths of both for more accurate and reliable predictions.
Uncertainty Quantification
Developing techniques to assess and manage uncertainty in machine learning models, enhancing their robustness and reliability in real-world deployments.
Dynamical Systems
Developing methods for analyzing, modelling, and forecasting complex dynamical systems, with applications in energy and environmental monitoring.
Explainable AI (XAI)
Developing machine learning models that are transparent and interpretable, allowing users to understand and trust the decisions made by these models.
Real-world impact first
We bridge the gap between cutting-edge ML research and industrial practice. Every project we undertake is driven by the goal of delivering measurable, practical impact for our partners.
Research Projects
AI4SWEng – AI-Driven Software Engineering
European research project investigating how large language models and AI agents can assist and automate software engineering tasks, from requirements analysis and code generation to testing and maintenance. SIMLab contributes expertise in hybrid modelling and uncertainty quantification for AI-assisted development pipelines.
Project website →CAPIA – AI-Based Cutting Tool Precision Control
Contrôle Autonome de la Précision des outils de coupe par Intelligence Artificielle. Innosuisse project with Eskenazi SA developing real-time machine-learning models for in-process monitoring and automatic correction of cutting-tool precision, reducing scrap rates and improving surface quality in high-precision machining.
Project website →JAXifer – Groundwater Level Forecasting
A JAX-based framework for 5-day groundwater level prediction from meteorological and weather forecast data. Uses differentiable hybrid models combining physics-based priors with data-driven components, enabling uncertainty-aware forecasts at regional scale.
GitHub →ML4HYDRO – Machine Learning for Hydroelectric Turbine Simulations
Domain-informed machine learning for the simulation of hydroelectric turbines. The project develops physics-constrained surrogate models that accurately replicate high-fidelity CFD simulations at a fraction of the computational cost, enabling rapid turbine optimisation and digital-twin applications for Swiss hydropower operators.
Sovereign Spearphishing Detection
Innosuisse project with Infomaniak developing a sovereign, open-source solution for automated spearphishing detection. The system combines large language models with behavioural analysis to identify highly targeted email attacks without relying on third-party cloud infrastructure, addressing privacy and data-sovereignty requirements for Swiss organisations.
TrunX – Domain-Informed Tree Growth and Mortality Modelling
SNSF Spark project developing domain-informed system-dynamics models of tree growth and mortality under changing climatic conditions. Combines differentiable mechanistic representations of carbon allocation and hydraulic failure with observational data to produce interpretable, uncertainty-aware forecasts of forest dynamics.
GitHub →Team
Prof. Dr. Gregory Mermoud
Gregory Mermoud is a professor at HES-SO Valais-Wallis and director of SIMLab. His research focuses on the intersection of physics-based modeling and machine learning, with an emphasis on developing interpretable and uncertainty-aware models for real-world engineering problems.
Dr. Cedric Travelletti
Cedric Travelletti is a senior scientist at SIMLab specialising in probabilistic machine learning and spatial statistics. His work addresses inverse problems in geosciences, with a focus on scalable Gaussian process methods and uncertainty quantification for large-scale environmental applications.
Glory Givi
Dr. Marc Gillioz
Marc is currently working on projects related to hydroelectric power production, in particular applying data analysis and machine learning techniques to:
- predict strain and fatigue for variable-speed turbines,
- detect anomalies in operational data for better maintenance planning.
At the HES-SO, Marc has also worked on problems related to power systems, such as power flow optimization through topological changes, modelling of hydroelectric production and high-voltage grids, and network reconstruction using Smart Meter data.
Marc's background is in high-energy physics, with a stint in software engineering.
Alexandre Veuthey
Alexandre Veuthey is a research engineer whose work at HES-SO and SIMLab focuses on the practical applications of Machine Learning for Computer Vision, particularly at the intersection of the 2D and 3D vision modalities. His prior experience in an industrial context enables data-driven solutions for vision tools based on cameras and other sensors.