GitHub Email Codementor LinkedIn

Portfolio

Selected Work.

claude-opus-4.6 reply

Recent work — a run of React Native mobile apps shipped solo to TestFlight and the App Store, plus production web platforms, AI experiments, and security-focused tools. →

CastLog

2026 · Founder & Engineer

CastLog

AI-powered fishing journal built for wet hands and flopping fish. Log a catch in under 15 seconds, identify species from a photo, and pull local regulations by GPS — fully offline. No paywalls on core features, no corporate data pipeline.

  • React Native
  • On-device AI
  • Offline-first
Case study

Problem

Existing fishing apps assume dry hands, full signal, and patience — none of which exist on a boat. Anglers either stop logging mid-trip or lose their data behind a paywall when they cancel.

Approach

Designed an offline-first journal with a 3-tap logging flow, on-device AI fish ID (94% accuracy), GPS-based regulation lookups, and a searchable knot library that works with no signal. Full data export (CSV/JSON) and private-by-default spots are baked in — users own their catch history.

Outcome

Now live on the App Store. Voice logging and Apple Watch support on the roadmap.

Live → App Store →
NutriLogix AI

2026 · Founder & Engineer

NutriLogix AI

AI-powered nutrition and fitness tracker. Snap a photo of any meal to log calories and macros instantly, scan barcodes for localized health scores, and sync activity in real time via Apple Health. Currently in TestFlight beta.

  • React Native
  • OpenAI API
  • Apple Health
Case study

Problem

Manually logging meals is tedious and inaccurate — users either skip it or spend minutes searching databases for food items. Existing trackers require too many taps and still get portion sizes wrong.

Approach

Built a vision model pipeline (OpenAI GPT-4o Vision) that identifies food, estimates portions, and returns structured macro data from a single photo. Barcode scanning adds localized health scores pulled from product databases. Apple Health sync keeps calorie burn and activity data in one place without manual input.

Outcome

Launched on TestFlight with early access users actively logging meals. Photo logging reduces manual entry time from ~2 minutes to under 10 seconds per meal.

NutriLogix AI architecture — vision and barcode pipelines merging into Supabase and Apple Health
Vision pipeline (camera → GPT-4o → macro parser) and barcode pipeline (scanner → product DB → health score) both feed into Supabase, syncing out to Apple Health.
Live → TestFlight →
VoiceLift

2026 · Founder & Engineer

VoiceLift

Voice-first workout logger. No forms, no friction — say your set and it's logged instantly. AI understands gym slang, supersets, and edge cases.

  • React Native
  • OpenAI API
  • Whisper
Case study

Problem

Form-based workout loggers break training flow — tapping through dropdowns between sets kills momentum and focus. Most lifters either stop logging mid-workout or don't log at all.

Approach

Built a voice-first pipeline using OpenAI Whisper for transcription and GPT-4o structured outputs to parse gym slang, supersets, and compound set descriptions into clean Supabase records. A confirmation sheet lets users verify before saving, with a ripple-animated voice button as the sole primary interaction. Stack: Expo SDK, Supabase (Postgres + RLS), RevenueCat subscriptions.

Outcome

Logging a set drops from ~15 taps to a single voice command. Currently in active development with early access at voicelift.app.

VoiceLift architecture diagram
Voice pipeline — mic → Whisper (STT) → GPT-4o Structured Outputs (gym-slang parser) → Set / Superset State Machine → Confirmation Sheet → Supabase · RevenueCat.
Live →
More on GitHub
Browse my full repository list and side experiments.
→
Let's work together
Have a project in mind? Let's talk.
→
View archive 37 more projects