Useful systems.
Measured outcomes.
I built three applications to explore financial filings, electricity-demand forecasting, and code review. Each one uses real public data where available and explains how its results are produced.
Three projects, with the details included.
Each project includes a demo, source notes, and an explanation of what the current version can and cannot do.
FilingIQ
Compare annual financial figures from SEC filings and trace each value back to its source.
GridCast
Fit a model to historical electricity demand and compare its forecasts with a simple seasonal baseline.
CodeReview Intelligence
Inspect public GitHub pull requests and see which changed lines trigger rule-based review warnings.
Make the decisions easy to follow.
I want the interface to be easy to use and the underlying process to be clear enough to inspect.
Keep the source visible
I link to public sources and label sample records or historical snapshots.
Compare against a baseline
I compare the model with a simple method to see whether the extra complexity helps.
Handle failures clearly
When an API is unavailable, the app explains the problem and uses a local example where possible.
Include the code and tests
The source archive includes setup instructions and tests so the work can be examined locally.