data-modeling

Build a MongoDB NoSQL analytics platform for Airbnb data with disaster recovery, high availability, and geo-distributed sharding

Build Python data transformation pipelines with a visual IDE and AI assistant

Scores a five-day forecast into a crop risk rating from deterministic per-crop thresholds. FastAPI, SQLAlchemy, React, Tailwind.

End-to-end European income distribution analytics pipeline using Eurostat data. Extracts and transforms data with Python and Pandas, models it in a star schema, loads it into PostgreSQL for SQL analysis, and delivers an interactive Power BI dashboard for exploring income trends across countries and demographics.

介護制度の公式資料を構造化し、実務の疑問から根拠へ辿る公開DB。制度理解・データ設計・独立照合・現行性と人手確認を分ける品質管理の個人プロジェクト。

Multi-tenant data and AI platform for ~40 SaaS customers: event ingestion, dbt daily marts, a Gemini AI analyst with fixed tools, and Postgres RLS tenant isolation proven by CI tests. Includes CSV self-serve upload, org deletion with receipt, cost controls and kill switches. FastAPI, Supabase, Render, GitHub Actions.

自治体のAI調達を公式一次資料から比較・構造化する個人研究。仕様・質疑・訂正・評価・公開結果を接続し、有効要件と公開証拠の限界を管理。

Linked Open Data Modeling Language

🛢️ Visualize and forecast oil & gas well production with a user-friendly dashboard for data cleaning, engineering, and machine learning training.

✈️ Build an end-to-end flight data analytics pipeline to process U.S. flight performance data, enabling insights into airline punctuality and efficiency.

Agentic AI for Data Vault 2.0 — requirements to contract-backed dbt code

[In progress] A comprehensive wrapper around Winnipeg Transit's API that integrates GTFS data to provide customized real-time bus location and service information, while maintaining historical records of buses and runs. The plan is to eventually integrate this application into my other transit-related projects.

Type annotations for specifying, validating, and serializing arrays with arbitrary backends in Pydantic (and beyond)

Point-in-time SEC fundamentals warehouse (dbt + DuckDB) that models financial restatements to eliminate lookahead bias

End-to-end data engineering project using Medallion Architecture, synthetic data and a Power BI Semantic Model.

SaaS growth analytics: activation, retention, churn and experimentation with SQL models, Python and Streamlit

dbt warehouse and long-stay prediction model on 173,000 Austin Animal Center records. Austin's live release rate held steady while dog adoption waits tripled; the model flags at intake which dogs and cats are likely to stay 30+ days. Tested end to end, with backtest monitoring and leakage checks in CI.

Hospital-operations analytics on 12,000 synthetic visits (FY 2024-25) — MySQL schema + KPI queries, seeded Python data generator, Power BI star schema with DAX. Revenue mix, admissions, LOS & discharge outcomes.

Analytics engineering reference project: 40 dbt models, 128 tests, SCD2 snapshots and margin-aware channel ROAS. Clone and run - builds on DuckDB with no credentials, verified end-to-end on Snowflake and Redshift. Optional Dagster orchestration, or integrate with your own Airflow.