Prototypes and tools designed and built on personal time, using Claude Code and Claude Design. Everything lives as code on GitHub — from single-file HTML tools to native iOS builds. Screens below are stills from the working prototypes — live walkthroughs available on request.
iOS health app concept. Reads bloodwork, sleep, and activity together and turns them into a daily plan. Includes a full onboarding flow, dashboard, logging, notifications, and light/dark theming.
AI health coach concept. Tracks recovery, training load, and supplements through conversation. Logs are staged and confirmed before anything is written, and a thread can be opened from any metric card with its context attached.
Eight lab panels from 2021–2026 consolidated into one dashboard: per-marker cards with status and plain-language context, trend charts, a hypothesis-driven read of symptoms, and a prioritized action plan.
Estimation tool for delivery teams: team composition, timelines, and cost scenarios, with an exportable client-ready report. Based on the estimation model developed at Technisys.
Sprint-level dashboard computed from raw input files — status, risks, and trends — with a shareable HTML report generated per sprint.
Live-scores companion for the 2026 World Cup: schedule search by team, group standings with per-team and per-group match links, and highlight access from any finished match.
Personal P&L computed from raw bank statements. Net-worth banner across personal, business, and portfolio views, budget tracking, spending analysis, and a built-in advisor panel.
Additional apps maintained as repositories on GitHub.
Training companion prototype — workout plans, logging, and progression. Will be rebranded before release.
Spanish-language wellbeing companion built for family. Personalized weekly guidance from a short onboarding, powered by the Claude API — private, on-device data.
Not everything is an app. Some of the most useful work is in how the AI itself is set up.
Persistent projects that convene a synthesized board of named specialists — a health advisory board spanning sleep, metabolic, and hormonal medicine, and a seven-advisor financial board grounded in Canadian tax rules. Each answers with evidence weighed across the pool and per-paragraph attribution, against my real baselines.
Scheduled agent sessions that organize the day: reading and triaging email, cleaning up the inbox, and preparing what needs attention — running as recurring tasks rather than one-off prompts.
Model Context Protocol connectors that give the assistant direct access to calendars, documents, and data sources — so analysis and automation work on live information instead of pasted snippets.