Working demonstrations
Most of Daniel's applied career was inside environments that cannot be shown. These pages take methods from that work and run them, live, in your browser on synthetic data invented for the page. Real algorithms, fake numbers, view-source welcome.
Telemetry Anomaly Lab
A live stream of synthetic sensor telemetry with three online detectors running against it (adaptive z-score, EWMA, CUSUM). Inject faults yourself — spikes, drift, stuck sensors, dropouts — and watch what each detector catches, how fast, and what it false-alarms on.
Run it →Entity Resolution Workbench
Twelve messy vendor records — typos, swapped fields, abbreviations — deduplicated in front of you with blocking, Jaro-Winkler similarity, and transitive clustering. Drag the threshold and watch clusters merge and split; paste your own records if you like.
Run it →Family Whole-Life System
A connected financial operating system on an invented household: double-entry ledger with editable rules and integrity checks, household/vehicle/asset registry with live net worth, 400-path Monte Carlo retirement projection driven by actual ledger spending, allocation glide-path check, DIME life-insurance adequacy, umbrella-liability trigger, and a scored estate-planning readiness checklist.
Run it →Workout Tracker
A functional training log with example data: log real sets, estimated one-rep-max trends, weekly volume, and automatic PR detection. Everything you enter stays in your own browser.
Run it →Manuscript Studio
Author-side book tooling on a sample chapter: content-addressed paragraphs (edit the text and watch the integrity registry catch it), typed claim extraction with a verification queue, and a live concept map.
Run it →Career Q&A (retrieval-grounded AI)
An LLM that answers questions about Daniel's background from a structured evidence database, with guardrails, grounding rules, and rate limits. The demo is that it exists and behaves; ask it something it shouldn't answer and watch it decline.
Ask it →Trust Calibration Lab
The behavioral-science demo: review ten pieces of AI work and accept or override each one. The page measures your own over-trust and under-trust — first with bare answers, then with calibrated confidence and the right to decline — and shows you against the anonymous average.
Measure yourself →Case studies
The longer stories behind each demo: the problem the pattern solves, how the method works, where the production version runs, and what it changed.
Read them →Résumé
The standard public résumé, rendered from the same structured career database that powers the assistant and the MCP server — readable here, downloadable as DOCX.
View / download →Live infrastructure status
Not a demo — the real monitoring surface for the systems that serve this site, including backup freshness and self-healing activity.
See it →Why synthetic data? Twenty-plus years of the real thing belongs to past employers and their security boundaries, and respecting that boundary is part of the demonstration.