Writing
Notes on statistics, Python, machine learning systems, LLM evaluation, and the places where implementation details change the conclusion.

A research-style account of a lavaan performance PR, using matched profiling runs to connect internal fast paths with measured latency and allocation reductions.
A rigorous tutorial on second-generation p-values, interval null hypotheses, frequentist interpretation, and applied R workflows with the sgpv package.

A JavaScript Leaflet walkthrough for joining GeoJSON boundaries to official 2024 population metrics and turning the result into an interactive choropleth.
systems
C++
A from-scratch C++ guide to dot products, transposes, matrix-vector multiply, cache locality, allocation control, blocking, SIMD-aware loops, and measurement discipline.

A refreshed version of my original Quarto note on building a Japan prefecture choropleth with Leaflet in R.
inference
STAT
A practical guide to using Bayes' rule, Beta priors, credible intervals, and posterior comparison for A/B tests.
systems
C++
A practitioner deep dive into latency measurement, cache locality, allocation avoidance, branch predictability, atomics, false sharing, syscalls, and p99 discipline.
inference
STAT
A practitioner deep dive into bootstrap aggregation, decorrelated trees, majority votes, OOB error, and why random forests work so well on tabular data.
inference
STAT
A compact note on using posterior predictive checks to catch quiet distribution shifts before model quality moves.
comparison
EVAL
Designing an evaluation loop that treats prompts, datasets, and model settings as versioned artifacts.

This post is a review of my short stint as a researcher in undergrad.

This post is a walkthrough of various power analyses in R.

This post is a walkthrough of a structural equation model in R.

The XGBoost Boosting Algorithm and an Application to the Forest Cover Type Dataset

My topic of my 2021 master's thesis during my final semester.

One of the most difficult projects I did in a statistics class during grad school.

A quick notebook on how to create and interpret naive bayes models.

