Case study
01 · The question
Challenge
Food matrices mix overlapping chemical signals, making quality monitoring and oil identification difficult with raw spectra alone.
02 · The work
Approach
I contributed Python pipelines combining spectral preprocessing, PCA, ratiometric analysis, feature engineering, NNLS decomposition, and classical machine-learning models.
03 · The value
Impact
The project demonstrates how accessible spectroscopy and interpretable AI can support practical food-quality monitoring.