I build the pipelines, warehouses and visualizations that turn raw data into answers — and lately, the agents that let you just ask.
I run everything linked here myself, from the server up — the sites, the REST API, the visualizations, the AI agents, four browser games, the analytics and the two warehouses behind them. It is pointed at what interests me most: data engineering, weather's effect on your body, sports data, good food, and a good game.
Equipping decision-makers with the tools — and now the plain-English interfaces — to make sense of the data.
Designing and running relational and non-relational stores — PostgreSQL, MySQL, SQLite, MongoDB. Ingesting and transforming JSON, CSV, Parquet, XML and YAML, and using Python, SQL and Bash to optimize each stage of the ETL and EtLT process. Pipelines run unattended on AWS Lambda and EventBridge, land in a PostgreSQL instance I manage myself, and back themselves up to S3 nightly with alerting when a run fails.
Explore the REST API docsHarnessing Python and its standard and third-party libraries — Pandas, Polars, PySpark, SQLAlchemy — to build robust data models. Applying advanced SQL techniques (CTEs, window functions, CASE expressions) to answer the hard questions, and using the relevant statistical and aggregation methods to make valid inferences from the results.
Putting a language model in front of a warehouse without handing it the keys. An LLM agent reads the live schema and writes the PostgreSQL; a Model Context Protocol server executes it as a SELECT-only role behind a query guard and a statement timeout. The web tier holds no database credentials and builds no SQL, and a token budget that fails closed keeps a public demo from becoming an open tab. The same pattern now runs over two warehouses, weather and sports, reusing one tested security model rather than inventing a second.
Try Ask WeatherShipping the work as software people can actually open. Flask, FastAPI, Streamlit, Dash and Bokeh applications, and plain-JavaScript games, self-hosted on a single EC2 instance behind nginx — TLS from Let's Encrypt and renewed on a timer, each app its own systemd unit, most with a dedicated service user and a memory ceiling. Every request is captured into PostgreSQL by an analytics pipeline and read back through a live dashboard, and every site is measured rather than assumed: Lighthouse audits, security-header grades, and Core Web Vitals from real visitors.
See the data visualizationsBoth warehouses are built the same way: validate the data on the way in, serve it read-only, then put a plain-English agent and a visual explorer on top.
Hourly observations from NOAA and Apple WeatherKit, loaded by unattended AWS pipelines into PostgreSQL on EC2.
Ask a weather question in plain English. Gemini writes the PostgreSQL, an MCP server runs it read-only against 9.3 million rows, and the query is shown beside the chart.
Apple WeatherKit forecasts loaded into a self-managed PostgreSQL database on AWS EC2 every four hours, orchestrated by Lambda and EventBridge.
Hourly observations for 112 US airport stations in a self-managed PostgreSQL instance on AWS EC2, refreshed nightly from NOAA's GHCNh parquet files by Lambda and EventBridge.
Historical observations and forecasts from the two PostgreSQL databases behind the visualizations, served as JSON with interactive documentation and an OpenAPI spec.
Every FBS college football and NFL play since 2014, from ESPN, one row per play with the people identified by stable id.
Ask about any college football or NFL play since 2014. Gemini writes the SQL, a read-only MCP server per league runs it, and the query is shown with the answer.
Every play, filtered by league, side of the ball, team, player and outcome. A Dash app reading a DuckDB snapshot, with drill-down and CSV or Excel export.
ESPN's feed parsed into two fact tables over shared dimensions in PostgreSQL, mirrored to EC2 and split into per-league serving schemas, rebuilt weekly in season.
Smaller builds, each on a different stack.
A typing-rhythm coach for writing AI prompts: burst speed, thinking pauses, backspacing and spelling, from every keystroke's timing.
The family recipe collection in Flask and PostgreSQL: search, print to PDF, or transcribe a recipe card from a photo with Claude or Gemini.
Interactive Bokeh apps, built in Python and served straight from the weather warehouse.
Forecasts from the Apple WeatherKit REST API, refreshed every four hours by an automated pipeline.
Explore
Historical and forecasted metrics side by side, to aid in discovery and analysis.
Explore
Utilizing the power of SQL window functions (LAG) to drive metric creation.
Explore
Rich historical data representations in a concise manner, to aid in pattern recognition.
ExploreI take on freelance data work, large or small: pipelines and warehousing, analytics and modeling, or putting a language model in front of data you already have.
