Key Capabilities
- Molecular Featurization at Scale – High-throughput molecular featurization pipelines for classical descriptors and deep learning embeddings enabling large-scale fuel property prediction.
- Predictive & Generative Modelling – State-of-the-art predictive models for property estimation and generative frameworks for designing novel sustainable fuel candidates.
- Inverse Fuel Design – Advanced inverse design strategies leveraging diffusion models, genetic algorithms, SciPy optimisation, and reinforcement learning to explore chemical space and optimise fuel formulations.
- Uncertainty Quantification – Integration of methods like EnbPI for robust prediction intervals to ensure reliable surrogate model predictions.
- Simulation Acceleration – ML surrogates to accelerate 0D and multidimensional combustion simulations while maintaining accuracy.
- Data Integration & Visualisation – Tools for combining experimental, simulation, and computational data with interactive visualisations to explore fuel performance efficiently.
