Agent skill for formatting and posting Slack messages: mrkdwn vs markdown, bold syntax by route, Block Kit rich_text, and a lint script. Works with Claude Code and Codex.

Profile README

Convert Japanese DPC inpatient data to the OMOP Common Data Model v5.4 (built from the public specification and synthetic data only)

Hierarchical Bayes choice-based conjoint analysis in Python (PyMC), with Sawtooth-style outputs and a market simulator

An agent skill for Claude Code, Codex and Cursor that rewrites AI-sounding Japanese into natural Japanese without changing the facts. 59-item checklist across six layers, plus an automated checker.

Fixed-income quant tutorials in Jupyter: every model implemented from scratch in NumPy/SciPy and cross-checked against QuantLib (curves, short-rate models, rate derivatives, risk)

Market volatility as a latent state: stochastic volatility and regime-switching HMMs in PyMC, with the forward algorithm verified against exact enumeration

VoC analysis: emotion trajectories from transcripts, plus annotation-quality vs outcome-weight analysis on Persuasion for Good (how good must annotation be for attribute weights to be decision-grade)

S/T/X meta-learners for uplift modeling, benchmarked on the Criteo uplift dataset with bootstrap confidence intervals on Qini and AUUC

How many customer segments? Dirichlet process mixtures decide from the data: CRP and collapsed Gibbs from scratch, applied to retail RFM features

Item response theory and ordinal regression in PyMC: Rasch, 2PL and graded response models on classic test data, plus why averaging Likert scales misleads

Sequential Bayesian A/B testing with expected-loss stopping rules: executed notebooks and a when-to-stop decision calculator (English/Japanese)

Rank anything from pairwise comparisons with uncertainty: Bradley-Terry, Davidson and Plackett-Luce in PyMC, applied to LLM arena votes, food preferences and tennis

Is your backtest overfit? CSCV/PBO, Deflated Sharpe Ratio and MinBTL implemented and verified in executed Jupyter notebooks

Classic GoF design patterns in modern, dependency-free Python — hands-on tutorials with runnable demos, tests, and exercises

Dark and dim themes for kintone apps with per-user switching and OS auto mode

Compare two kintone records side by side with field-level and inline text diff

Assign new kintone records to team members in rotation or by current workload

Track how long kintone records spend in each process status and locate bottlenecks

Generate realistic Japanese test data for kintone apps, reproducible with a fixed seed