The work behind the release.

Selected product and engineering work, with my contribution and available evidence.

Product DevelopmentW–2

Personal bookshelf in Paguely — Product Development

Book discovery in Paguely — Product Development

W–01 · Consumer reading app, my own product

Paguely: a book app built and shipped alone

Choosing the next book is hard when every cover pushes you somewhere. Paguely offers a blind date with a book: a mood quiz gives clues, and the title stays sealed until the reader commits.

My role: Founder and sole engineer
Period: July 2026 – present
What I built: Flutter app with Clean Architecture by feature, BLoC and go_router.
Supabase backend with 24 migrations and 13 edge functions.
RevenueCat subscriptions with a webhook, Sign in with Apple and Google, an iOS home-screen widget.
Book recommendations generated with OpenAI.
The marketing site and a privacy page written from an audit of the real code.
Stack: Flutter, Supabase, RevenueCat, Firebase, PostHog, OpenAI
Result: Live on the App Store and Google Play.
URL: paguely.app

ScalePilot agency website — Product Development

W–02 · Marketing agencies, my own product

ScalePilot: landing pages from a guided brief

Agencies spend days on landing pages that follow the same pattern.

My role: Builder
Period: June 2026 – present
What I built: A guided brief, then a pipeline that builds, checks and publishes a designed page.
Multi-tenant setup with immutable site versions, rollback, custom domains, billing and teams.
Stack: Node, Anthropic SDK, Playwright, Tailwind, GSAP, Supabase, Vercel, Railway
Result: Live as a multi-tenant service. 99 commits between June and October 2026.

Software DevelopmentW–3

meinunfallauto vehicle marketplace — meinunfallauto

W–03 · meinunfallauto · Automotive commerce, Germany first

meinunfallauto: both halves of an EU automotive commerce platform

A platform for damaged, accident and regular vehicles needed an admin and partner backend, a fast multilingual customer site, deposits on reserved cars and data imported from an old system.

My role: Sole engineer, backend and web
Period: May 2026 – present
What I built: Laravel 12 backend with admin and partner dashboards, two-factor sign-in, PDF invoices and background jobs.
Next.js 16 customer site, German by default, with English and Arabic right-to-left.
Stripe reservation deposits tied to inventory holds, retry-safe webhooks and real refunds.
A domain-specific search engine with its own lexicon and query parser.
Legacy database import over a read-only connection; an in-admin CMS with a cached content API.
Stack: Laravel 12, PHP 8.2, Next.js 16, React 19, Tailwind 4, next-intl, Stripe, Playwright
Result: Both codebases written by me: about 87 backend and 90–95 frontend commits (October 2026).

W–05 · Consumer dating app, European startup

A production dating app, mobile owned end to end

A dating app for iOS and Android needed one engineer to own the whole mobile side: features, payments on both stores, chat that never loses messages, and releases with urgent fixes.

My role: Sole mobile engineer, alongside a backend team of about twelve
Period: October 2025 – present
What I built: In-app purchases with StoreKit and Google Play Billing 8: receipt verification, acknowledgement, entitlement polling, cross-account ownership conflicts and Google real-time developer notifications.
Rebuilt chat pagination and delta reconciliation, removing permanent message gaps caused by a server-side cap.
The API layer extracted into a versioned Dart package, so one commit reproduces a code-push-patched build exactly.
GitHub Actions quality gates and Codemagic signed builds per flavor, with Shorebird code push.
Platform work: state restoration, custom haptics with Android fallbacks, ML Kit face verification, image paste in chat.
Stack: Flutter, BLoC, go_router, dio, Socket.IO, drift, Firebase, Shorebird, Codemagic, GitHub Actions
Result: 1,090+ commits, 99% of the history; about 124k lines of Dart covered by about 112k lines of tests in 464 test files (October 2026).

Clematis onboarding interface with explicitly fictional demonstration business data. — Software Development

W–06 · Small businesses, DACH market

Clematis: multilingual web and customer onboarding

Connect multilingual service discovery with customer onboarding and operational tools.

My role: Web development and integrations
Period: Project work
What I built: Interfaces in Bosnian, German and English.
Payments, transactional email and CRM integrations.
Stack: Next.js, React, Stripe, Resend
Result: Implemented the journey from service discovery through customer onboarding.

ManyRoads public product visual showing route discovery. — ManyRoads

W–07 · ManyRoads · Outdoor and travel, multilingual walking and hiking routes

ManyRoads and ManyQuests: two route apps on one Flutter codebase

A walking and hiking platform needed two apps from one codebase, with maps and several languages.

My role: Flutter developer
Period: January 2026 – present
What I built: Features across both apps on a shared codebase with three build flavors.
Mapbox maps, localisation, product analytics, Codemagic builds.
Stack: Flutter, async_redux, go_router, Mapbox, easy_localization, PostHog, Codemagic
Result: 107 of 224 commits in the ManyRoads app (January–October 2026).

W–08 · Consumer marketplace for city services

Features for a large city services marketplace app

A large consumer Flutter app needed new features ported from Figma, map work and localisation, without breaking a codebase shared by eight developers.

My role: Flutter developer in a team of eight
Period: July 2026 – present
What I built: New profile and guided-tour features ported from Figma and checked against the designs.
Map clustering work in the app's Google Maps package.
Localisation and design-conformance fixes across the app.
Stack: Flutter, Clean Architecture, BLoC and Riverpod, Firebase, Google Maps, Stripe
Result: Work merged into a production app with roughly 2,000 Dart files and 260+ test files (October 2026).

W–10 · Basket · Retail, shopping and cashback, United Kingdom

Basket: lead Flutter work on a UK shopping and cashback app

A shopping and cashback app needed lower Firestore cost and load, new engagement features and a navigation system that kept links working.

My role: Lead Flutter Developer
Period: August 2025 – June 2026
What I built: Batched Firestore queries that cut active listeners by about 97%.
Virtual try-on with ML Kit image-quality checks, sharing and watermarking.
Price-drop and back-in-stock notifications, local deals, deep links.
A navigation refactor across 48+ routes that keeps query parameters through redirects.
RevenueCat subscriptions; Codemagic CI/CD with dev, staging and production flavors.
Stack: Flutter, Firebase, RevenueCat, GoRouter, Codemagic
Result: Over 99% crash-free sessions; slow traces down 35%.

Apini public App Store screen showing laboratory records. — ApiniLabs

Apini public App Store screen showing a honeycomb laboratory notebook. — ApiniLabs

W–11 · ApiniLabs · Laboratory management, Switzerland

ApiniLabs: real-time sync and lab hardware for a Swiss lab-management platform

Lab teams needed the same data on several devices at once, without conflicts or slow screens, and the app had to talk to the instruments on the bench.

My role: Lead Flutter Developer
Period: June 2024 – August 2025
What I built: Real-time multi-client sync on Firestore and Cloud Functions.
Bluetooth Low Energy connection to an Ohaus Explorer precision analytical balance.
Native SDK integration of a hardware barcode and material scanner and a label printer, through Kotlin and Swift platform channels.
Crash and performance monitoring with Crashlytics and Performance Monitoring.
Stack: Flutter, Firestore, Cloud Functions, flutter_blue_plus, Kotlin, Swift, Crashlytics
Result: Data performance up 30%; over 99.5% crash-free; slow traces down 35%. The app, Apini, is on the App Store.

Concept illustration of sealed film discovery for Sealed Reel. — Software Development

W–13 · Consumer film and series app, my own product

Sealed Reel: a film app built with my agent system

Choosing what to watch tonight takes too long. Sealed Reel answers with a mood quiz and a sealed "blind date" title that stays hidden until you open it.

My role: Founder; built with my multi-agent delivery system
Period: October 2026 – present
What I built: A vault, a watchlist, series tracking, a daily drop and a taste profile.
A Pro subscription with RevenueCat; Supabase with row-level security on every table.
Local core and design packages, custom shaders, a home-screen widget and GitHub Actions quality gates.
Stack: Flutter, Riverpod, go_router, Drift, freezed, Supabase, RevenueCat, Firebase
Result: Not released yet. 445 Dart files, 314 test files and 2,027 passing tests (October 2026); most of the code came from my agent delivery system.

W–14 · Stage Coding · Enterprise time tracking, Bosnia and Herzegovina

Stage Coding: an offline-first enterprise time-tracking app

An enterprise time-tracking app had to keep working without a connection and sync every entry to the backend without losing or duplicating time.

My role: Lead Flutter Developer
Period: September 2023 – June 2024
What I built: An offline-first Flutter app that records time on the device and syncs with a Node.js backend.
Code standards for the team, and mentoring for two junior developers in my first lead role.
Stack: Flutter, Node.js
Result: Zero sync issues in production.

W–15 · Tech387 · Hospitality, fitness, social and productivity apps

Tech387: six-plus Flutter apps across hospitality, fitness and productivity

A Sarajevo software company needed Flutter apps for clients in several industries, each with sign-in, payments and releases on both stores.

My role: Flutter Developer
Period: June 2021 – September 2023
What I built: Six-plus apps released to the App Store and Google Play.
AI workout generation, biometric login, a media pipeline and subscriptions.
Payments with Stripe and Monri, a Bosnian payment gateway.
Backends on Firebase and AWS (AppSync, Cognito, S3).
Stack: Flutter, Firebase, AWS AppSync, Cognito, S3, Stripe, Monri
Result: Over 99% crash-free; performance up to 60% better through lazy loading, background sync and caching.

AI ImplementationW–4

W–04 · Software delivery, my own tooling

A multi-agent delivery system on Claude Code

I wanted agents to take a product spec to a tested codebase, with roles like a real team: discovery, design, engineering, QA, release.

My role: Designer and builder
Period: 2026 – present
What I built: 100+ role-specialised agents and workflows, with judgement roles on a stronger model and the rest on faster ones.
A code audit workflow where every finding is challenged by three independent reviewers before it is reported.
An App Store Connect API client and a local dashboard to watch runs.
Stack: Claude Code, Claude API, TypeScript, Flutter, Supabase
Result: One run produced a 445-file app with 2,027 passing tests. The first full run showed that healthy code is not a working product: a new user could not finish onboarding. Verification was rebuilt around user journeys, and the next run caught the broken path first.

Concept illustration of a reviewed content-production workflow for AutoReach. — AI Implementation

W–12 · Marketing automation, my own tooling

AutoReach: a content engine for short video and posts

Paguely, my book app, needs a steady flow of short videos and posts on TikTok, Instagram, YouTube Shorts and Pinterest, plus an SEO blog, without a person producing each one.

My role: Designer and builder
Period: September 2026 – present
What I built: A loop that plans, generates, checks, publishes, measures and learns.
Versioned prompts with evals, and three different models for generation, scoring and audit.
Official platform APIs only.
A console and a worker in a Turborepo monorepo with nine packages, with zod validation at every boundary.
Stack: TypeScript, Node 22, pnpm, Turborepo, Postgres, MinIO, zod
Result: In development: 135 commits and about 930 TypeScript files between 29 September and 11 October 2026.

Trading botsW–5

W–09 · Retail trading tools, client project

A Telegram-to-MT5 signal copier

Signals posted in several Telegram groups had to reach several MT5 accounts quickly, with a way to pause everything at once.

My role: Builder
Period: 2026
What I built: A Python service that reads Telegram signals and places orders on multiple MT5 accounts.
Webhooks between components.
A STOP file that halts trading without killing the process.
Stack: Python, MetaTrader 5, Telegram
Result: Finished and delivered to the client as specified. No trading results are claimed.

Trading carries a high risk of loss. Trading risk disclosure

Check sheetW–6

What needs to work next?

CS-01

Tell me what exists today, the problem you want to solve and the next milestone. We can start with a focused review, a product release or an AI pilot.

What the product does and who uses it.
What exists already: designs, code, a backend, a store account.
Your deadline, and anything fixed such as a launch date or a platform.

Discuss your project hello@merimtufo.com