MATLAB.

Where MATLAB fits on a long engagement.

MATLAB estates are usually research assets: models developed over years by engineers or quantitative researchers, validated against physical systems or market data, and trusted precisely because they have been used for a long time. They are rarely written as production software, because that was not the purpose.

The tension arrives when a model needs to run in production. Rewriting in Python risks losing behaviour that was never fully specified anywhere but the MATLAB code; keeping MATLAB in production means licensing and an integration path. Neither answer is automatically right, and the decision needs the validation cost accounted for honestly.

What an assigned team does with MATLAB.

The safest route is usually a rewrite with rigorous equivalence testing against the original across a wide input range, rather than either a blind port or an assumption that the two will agree.

That testing is the bulk of the work and the part most often underestimated, which is why it is scoped explicitly as part of how the monthly fee is built.

What we use MATLAB for.

• Existing models understood before moving Behaviour documented, because the code is often the only specification.
• Equivalence testing across the input range A rewrite verified against the original rather than assumed to match.
• Integration where a rewrite is not justified MATLAB called from a production system, where porting cost exceeds the benefit.

How MATLAB capacity is assigned.

MATLAB work is assigned under managed data services, with equivalence validation scoped as the main deliverable in any migration.