top of page

Growth through Mathematical Precision

When your simulations are too slow, your models aren't accurate enough, or your pipelines can't scale, standard tools run out of road. I work with companies at exactly that point, applying rigorous numerical methods and scientific computing to problems that genuinely require them.

Solver Optimisation

Your simulation works, but it is too slow or too expensive to run repeatedly. I use advanced numerical methods, including domain decomposition and preconditioning, to reduce computational cost while preserving accuracy.

Optimisation & Decision Support

Complex operational problems often involve limited resources, competing objectives and hard constraints. I develop mathematical optimisation models for planning, allocation and operational decision-making.

Numerical Software Dev

I don’t just advise, I deliver. I design and implement robust numerical software in Python, FEniCS, BEMpp and C++, with an emphasis on validation, maintainability and integration with existing workflows

Coupling Complex Systems

Real systems often involve several models, domains or physical processes. I develop computational frameworks that allow these components to interact efficiently and consistently.

Removing Manual Tuning

Many numerical pipelines rely on parameters that someone has to set by hand. I design solvers that find optimal values automatically, eliminating trial-and-error and making your system reliable without expert intervention.

Engagement Models

I offer two main routes:

-direct consultancy for defined technical challenges;

- longer-term research and development partnerships, including collaborative grant or Knowledge Transfer Partnership opportunities where appropriate.

Case studies

Design-to-simulation for manufacturing

A simulation process that previously took hours was reduced to minutes, a greater than 10x improvement in runtime, without any loss in accuracy. This was achieved by applying fast domain decomposition solvers to an isogeometric analysis workflow, exploiting the mathematical structure of the problem rather than simply adding computational power. The same methodology applies to any CAD-integrated or simulation-heavy workflow facing similar bottlenecks.

Not sure if your problem is the right fit?
Describe it to me, and I'll tell you honestly whether I can help. Get in touch →

​Dr MichaƂ Bosy

School of Computer Science and  Mathematics

Kingston University London

Penrhyn Road

KT1 2EE Kingston upon Thames

​

​© Michal Bosy 2026

Open to industry collaboration
and academic partnerships

  • google-scholar
  • LinkedIn
  • researchgate
  • orcid
bottom of page