Insights
Where spectroscopic methods actually fail.
Short technical notes on method design, substrate and matrix effects, and leakage-free validation — the parts that decide the result long before the instrument does.
- chemometricsquality control
From R&D to release testing: chemometric models that survive an audit
A chemometric model that works in the lab is not yet a QC method. What it takes to move Raman and SERS models into regulated quality control, from validation to data integrity to lifecycle.
Read → - chemometricsmachine learning
Why your spectral classifier is probably overfitting, and how to prove it isn't
Most reported accuracies for Raman and SERS classification are optimistic. The cause is usually validation design, not the model. Here is how leakage creeps in and how to shut it out.
Read → - SERSnanomaterials
In SERS, the substrate and the matrix decide the result, not the spectrometer
Teams often invest in a better Raman instrument when their SERS problem is really a nanostructure and matrix problem. Where enhancement actually comes from, and what that means for method design.
Read →
Recognise one of these failure modes in your own data?
If a classifier that shone in development is falling apart on new batches, the fix is usually upstream. I can help you find it.