How we build things, in detail, with the code. This is not marketing — it is the evidence for whoever does your technical due diligence.
Twenty-four worked articles across three knowledge bases, plus the technical writing from our Insights archive.
How we build things, in detail
Every article is the working notes from real client or research work, kept as written rather than tidied for publication. This is the evidence for whoever does your technical due diligence.
Sixteen worked steps in R, from loading data and one-by-one variable analysis through PCA, monotonicity testing, outlier analysis and Gaussian copula encoding. Real code, runnable.
Eight steps: semantic tagging via LSEG-PermID, principal component analysis, k-means on factor scores, cluster statistics, and evaluating fairness of a selection process. Published with a live GitHub repository.
The DLMM internal rating model, built and explained end to end.
Why grounding a model in your own documents beats fine-tuning for enterprise use, and how the retrieval-augmentation-generation loop actually works.
Turning unstructured commentary into something you can count, test and report.
Closing the data gap when you do not have enough labelled examples to train on.
Where a rules engine outperforms a model, and how to tell which you need.
Written by Christophe Monget and the MoiraCorp technical team.
