These articles focus on the parts of machine learning that determine whether a model remains useful in production: temporal validation, feature design, anomaly detection, probability quality, explainability, deployment, and the engineering around the model.
Why Payrithm does not simply sort invoices by late-payment probability, and how I combine calibrated ML risk with value, urgency and collection activity.
Why late-payment prediction needs more than a chronological train/test split, and how I reconstruct what was actually knowable when each Payrithm prediction would have been made.
How I designed Payrithm’s machine-learning pipeline to predict late invoice payments without data leakage, evaluate probability quality, and turn model outputs into actionable collection priorities.
The often-overlooked craft of transforming raw data into informative features — covering encoding, scaling, missing values, interactions, and selection.
How to build a robust anomaly detection pipeline for multivariate sensor data using LSTM autoencoders in PyTorch, with practical tips on thresholding and deployment.
I use Google Analytics to understand how visitors use this portfolio and improve it. Analytics cookies and page-view tracking remain disabled unless you accept. You can change your choice later from the footer.