I build mobile apps, games, and full-stack products end to end, from architecture through shipping. Comfortable across React/Next.js, TypeScript, Flutter, and Dart on the client; Node, Deno, Firebase, and Supabase on the backend; and multi-provider LLM pipelines (OpenAI, Anthropic, xAI) for AI features. Shipped apps on the App Store and Google Play, wired Stripe and RevenueCat billing, and designed HIPAA and COPPA-oriented products.
Building software end to end, from architecture to release
Full-stack product builder shipping SaaS platforms, AI tooling, mobile apps, and games. I work from architecture through release, with a bias toward maintainability, clear documentation, and operational reliability.
End-to-End Product Ownership
I handle discovery, architecture, implementation, testing, and iteration from idea to production-ready delivery.
AI-Native Full-Stack Systems
I build across React/Next.js, TypeScript/Node, Flutter, Firebase, and LLM-driven workflows for practical user outcomes.
Production Discipline
I prioritize maintainable architecture, clear documentation, testing practices, and operational reliability.
Projects
Product Portfolio
Breakdown of every product I have built across SaaS, AI, mobile, and game systems, with explicit deployment status.
6 Live1 Not Deployed
Amora (Beacon)
Social Platform (Dating, Friends, Groups)
Not Deployed
Social app blending dating, friends, and groups in one Flutter experience, with matching, chat, and moderation systems.
Riverpod state with H3 hex-based proximity matching, roughly 45 Cloud Functions covering moderation, rate limiting, sanctions, and user tiering, and over 20 Firestore collections governed by 24 KB of security rules. Crashlytics, Analytics, Remote Config, and FCM wired end to end.
Cross-platform turn-based strategy game with a rules-driven engine, production-wired multiplayer, and a minimax AI opponent.
Riverpod rules engine encoding the full PRELADMO ruleset on a 5x5 board with layered tie-breakers, Firebase Realtime Database rooms with bot-fallback matchmaking, and a minimax AI with difficulty tiers running in background isolates. Signed Android builds, Crashlytics, Performance, RevenueCat, and AdMob are wired for both stores.
Flutter arcade shooter with a custom 60Hz game loop, campaign-to-endless progression, boss practice, and server-validated ranked leaderboards.
Riverpod state across 38 gameplay modules covering simulation, bosses, and powerups, backed by Firebase Auth, Firestore, and Cloud Functions that own all score writes. AdMob interstitials with a cadence service and RevenueCat IAP drive monetization, Remote Config drives forced-update gating and store URLs.
Flutter and Flame idle-merge creature collector with 17 species and a dozen Flame-powered minigames, shipped on both stores.
Firebase-backed accounts, cloud save, leaderboards, and live events, with a live-ops layer spanning battle pass, daily rewards, referrals, and minigames across rhythm, runner, stacker, and puzzle genres. Google Mobile Ads and RevenueCat handle monetization, Crashlytics, Analytics, and Remote Config round out the runtime.
Venture studio website showcasing portfolio companies alongside a multi-entry founder pitch funnel and server-side abuse controls. RAVEN Novel Engine (below) is one of the companies in the portfolio.
Zod-validated pitch funnel across four entry points, routed through a Supabase edge function that applies per-IP rate limiting (SHA-256 hashed IP, 3 per hour) and a honeypot field before dispatching to Resend. Pitches are emailed, never persisted, which keeps the site effectively stateless.
Multi-provider AI novel platform (OpenAI, Anthropic, xAI) running on a Supabase backend for auth, Postgres with row-level security, private image storage, and Deno Edge Functions that power the story generation pipeline. Guided wizard, durable chapter-by-chapter generation, continuity via story blueprints and bibles, Stripe billing, and an extensive security and COPPA audit trail. Linked from MRK as a portfolio company at ravenmagic.ai.
Open-source TypeScript browser agent that drives a real browser to scrape product signals from Reddit, Hacker News, Product Hunt, and IndieHackers, tailor resumes against live LinkedIn job postings, and run multi-step app-discovery pipelines. Every complex workflow ships as a self-contained multi-agent prompt pack.
Playwright-driven command-loop core (registry, event bus, state tracker, rate limiter, WebSocket remote-control plugin). Each target surface (LinkedIn, Reddit, Hacker News, Product Hunt, IndieHackers, app discovery, generic scraping, blackjack) is its own plug-in domain module. LLM calls route through a shared OpenAI and Anthropic client. Three prompt packs totaling 30+ numbered agent-assignment files coordinate the longer pipelines (resume tailoring, app discovery, Reddit analysis) end to end.
End-to-end scheduling SaaS across four coordinated builds on a shared Firebase project and Firestore schema.
A cross-platform Flutter booking app with Stripe payments, Google and Outlook calendar sync, and HIPAA-oriented security rules. A React CRM for leads, pipelines, sequences, forecasting, and reporting. A React admin console for billing, RBAC, onboarding, and operations. A static marketing site at empoweredbyai.net that routes traffic out to the three portals. Over 5,400 lines of Firestore rules enforce the shared auth and data model.