I contribute to open-source machine learning, statistics, and forecasting libraries, and to low-level systems and command-line tooling. My contributions are small, well-tested correctness fixes: numerical accuracy and API-contract fixes in the Python data-science ecosystem, and GNU parity for Rust CLI utilities.
- 🔭 Recent focus: correctness and scikit-learn API-contract fixes across the ML / Kaggle ecosystem.
- 🧰 Languages: Python, Rust, TypeScript / JavaScript.
- 🧠 Interests: time-series forecasting, computational statistics, anomaly detection, dimensionality reduction, LLM / AI tooling, and parser / CLI correctness.
- 🎓 Background: MSc Computer Science, BSc Economics (ITAM).
Machine learning and data science
- dmlc/xgboost — fixed the
plot_importancevalue-label offset for small importances (#12497) - koaning/scikit-lego — bounded the
IntervalEncoderaveraging window on both sides (#819) - sktime/sktime —
mean_squared_log_errorcleanup (#10816) - unit8co/darts — corrected the
extract_subseriesreturn contract (#3184) - tslearn-team/tslearn — enforced the LCSS Sakoe-Chiba constraint (#705)
Systems and CLI (uutils, GNU parity)
- uutils/coreutils — fixes across
uniq,fold,join,numfmt,pr,unexpand, andprintf - uutils/findutils — fixes across non-UTF-8 argv,
-size,-maxdepth,-newerXt, and-type
More contributions in review across pyod, feature-engine, category-encoders, mlxtend, pingouin, prince, tsfresh, umap-learn, and torchmetrics.
🤝 Open to interesting open-source collaborations in ML, forecasting, and systems tooling.