Daniel Wlazło¶
Data Science Manager at Allegro Pay — trustworthy machine learning for credit risk, in Warsaw.
About¶
At Allegro Pay I lead the modelling work behind one of Central Europe’s largest consumer‑finance platforms. My focus is application scoring — PD models for individual borrowers — and, increasingly, what trustworthy machine learning looks like in a regulated domain: fairness, calibration, explanation.
I’ve spent close to a decade in credit risk and data science, across Allegro Pay, Hexaware (for AXA), PKO Bank Polski, and ING Bank Śląski. Before that I was a robotics engineer at Samsung and a UAV constructor at Aviation Technik. Robotics was where I started — it’s still how I think about machine learning: a system you can describe, debug, and put under load. Outside work I learn Czech, follow Formula 1, and tinker with robotics.
Notes¶
August 2026treecf: counterfactuals you can act on
The only honest answer to “what would it take to be approved?” is a counterfactual — constrained, verified, fast.July 2026Why SHAP values are not points
A SHAP waterfall looks like a scorecard. Read it like one and it hands out advice that is confidently wrong.May 2026Discriminatory uncertainty
A model can discriminate not only through its decisions, but through how confident it is in them.
Open source¶
probcal¶
Post‑hoc probability calibration for binary classifiers, built for regulated PD models. Three things scikit‑learn will not give you: logit‑scale diagnostics that keep a 1% default rate readable, a central‑tendency adjustment that ships as an auditable offset stage rather than a silent refit, and per‑grade regulatory backtests (binomial, Jeffreys). Around them: the calibration literature behind one API, exact inverse maps from policy cutoffs to raw scores, anytime‑valid monitoring, and JSON serialization. numpy is the only runtime dependency.
Documentation · PyPI · GitHub
treecf¶
Counterfactual explanations for tree ensembles (XGBoost, LightGBM, CatBoost, sklearn): the smallest feasible change that moves a declined application under the cutoff, respecting the constraints you declare. Unlike sampling‑based tools it can prove that a plan is the cheapest available, or that none exists, and every answer is verified against the parsed model.
Documentation · PyPI · GitHub
Earlier and experimental: concept-graph-xai (concept‑level feature usage and ablation), triadxai (three‑channel decomposition of credit scores), flaggam (interpretable GAMs built from univariate flags), swift (SHAP‑weighted drift monitoring).
Academic work¶
Doctoral thesis — Trustworthy credit risk under the AI Act¶
Supervised by Prof. Aneta Ptak‑Chmielewska. The thesis develops an integrated methodology that combines algorithmic fairness, uncertainty quantification, and interpretability, organised around a concept I call discriminatory uncertainty: whether a model’s confidence, and not only its decisions, is distributed fairly across groups. The core idea is set out in the note Discriminatory uncertainty.
Conference talks¶
Slides and materials: github.com/wlazlod/conference-talks
- The Scorecard Illusion: Why SHAP values are not points (and how they can fool you) — Data Science Summit, AI Edition 2026, Warsaw.
Why a SHAP value behaves like a contribution, not a coordinate — and how reading it as a point misleads scorecard interpretation. Write-up. - Curing Semantic Debt: Practical Applications of Knowledge Graphs in ML — GHOST Day 2026, Poznań, with Kamil Golis.
How knowledge graphs pay down the semantic debt that accumulates across ML pipelines — a practical, applied walk‑through. - Poza Feature Store: jak Ontologie Semantyczne mogą przyspieszyć rozwój modeli ML — Data Science Summit 2025, Warsaw.
How semantic ontologies can move ML development past the feature store, giving models a shared, reusable vocabulary.
BibTeX entries for all talks: conference-talks README.
Teaching¶
Teaching materials: github.com/wlazlod/didactics
Data Mining — lecture course¶
A finance‑first, problem‑driven Data Mining course for the SMMD‑ADA and SMMD‑AAB Master’s programmes (elective for three more). It follows CRISP‑DM through one retail‑credit problem: exploratory analysis and missing data; regression re‑framed from inference to prediction, with logistic regression as the credit‑scoring workhorse; trees, ensembles and honest evaluation, including how leakage detonates; clustering and PCA for segmentation; association rules and text mining on complaints; neural networks and responsible ML.
Course script (HTML) · Repository
Research Workshop — industry project with Allegro Pay¶
Industry‑mentored project for the Research Workshop course at the Faculty of Mathematics and Information Science. Three companies contribute a project each year and guide the student team through it; I co‑prepare the brief and mentor the team. 2025: conformal prediction for consumer‑credit scoring. 2026: counterfactual explanations in credit‑risk modelling.