· MMXXVI
AlexMatosOlive
Computer Engineering student at UB and Co-founder & CTO at WRDB, building real product end to end. A native AI app in TestFlight beta, heading to the App Store, serverless backends, and web.
An AI wardrobe, heading to the App Store.
AI-powered virtual wardrobe, native on iOS. As Co-founder and CTO I own it end to end: the app, the backend, the AI, and the web.
What it is
A B2C native iOS app powered by AI, with personalized outfit recommendations and closet organization. In TestFlight beta with real users, heading to the App Store.
How it's built
Swift and SwiftUI following Clean Architecture (Domain → UseCases → Data → Presentation), on a serverless backend with Supabase and Cloudflare Workers.
Applied AI
Computer-vision models for clothing recognition and styling, a product site in Next.js, and an AI-assisted workflow built with Claude Code and MCP.
A labor law, turned into software people trust.
A workforce time-tracking and compliance platform for Spanish small businesses, built around the mandatory workday-registration law. In production, used daily by a real company.
What it is
Spain requires every company to log working hours, and most small businesses still do it on paper. EasyFichi lets a shop clock people in from a phone, a counter tablet, or the web, and stay on the right side of the law.
How it's built
One Flutter codebase becomes an admin app, an employee app, and a kiosk, across web, iOS, and Android, in four languages. Behind it sit Node.js Cloud Functions over Firestore, a modular compliance engine, and Claude parsing the documents that arrive by email.
The hard part
Anyone can build a clock-in screen. The real work was making the records defensible in a labor inspection: SHA-256 hash-chained audit trails, and a risk radar that flags what is not compliant before an inspector does.
Walking audio tours, built by an AI pipeline.
Self-guided walking audio tours through a city, narrated point by point and fully offline. The app is Flutter; the real product is the pipeline that writes the tours.
What it is
You pick a neighborhood, start walking, and Citourfy narrates each point of interest as you reach it. You'll buy a tour once and it never expires; it downloads in full so it works with no signal.
The pipeline is the product
The tours are not written by hand. A nine-step pipeline discovers points from OpenStreetMap, Wikipedia, and Google Places, has an AI generate the narration, computes real walking routes, and packs it into versioned bundles on Firebase.
Cheap, fast, and honest
The hard part was an AI pipeline that scales without hallucinating. Every model sits behind one interface with cost logging, bulk work runs through a batch API at a fraction of the price, and every fact is anchored to a real source before it ships.
Stack
- Flutter
- Firebase
- Cloud Functions
- Claude API
- React + Vite
- Google Places / OSM
Track migraines. Understand the patterns.
A migraine tracker for iOS and Android that turns detailed episode logs into charts, reports, and clinical alerts. A full Flutter rewrite of an earlier native iOS app, on the same backend.
Understand, don't just log
People record every episode in detail: type, pain, triggers, symptoms, and what actually helped. MyGrain turns that history into patterns you can act on, not just a diary.
One backend, two apps, zero migration
It is a Flutter rewrite of an existing iOS app, and current users could not lose anything. So it keeps the same Firebase backend and reads the old document shapes, with strict Clean Architecture and an offline queue so a save never fails silently.
Where it gets interesting
The parts I am proud of are hidden: ICHD-3 chronic classification, a medication-overuse alert over a rolling window, a weather correlation engine with no API key, and GDPR deletion that wipes every trace the moment an account is removed.
Stack
- Flutter
- Dart
- Riverpod
- Firebase
- Firestore
- Hive
You built the product. Let the machine market it.
An AI marketing machine for solo founders. Describe your brand once, and it plans, writes, and auto-publishes a full week of on-brand Instagram content. Built in a sprint at the Base Jump accelerator in Barcelona.
What it is
Founders can build but rarely have time to market. BuildLoud turns a 30-second brand description into a week of Instagram content: carousels, reels, and posts, published on autopilot so shipping code never stops for a content calendar.
How it's built
Next.js and TypeScript on Supabase. Claude plans the week and writes every post, Higgsfield and fal render the founder's face consistently across all of it, and the Instagram API publishes on schedule. Stripe handles billing, and a credit meter prices every generation.
Built in a sprint
Made with one teammate at Base Jump, a v3v-ventures-backed accelerator in Barcelona. The part I like most is the weekly loop: every Sunday it reads saves, reach, and comments, then reweights the next week around the hooks that actually landed. Face-locked generation keeps it from looking like generic AI.
I'm a Computer Engineering student at the University of Barcelona and co-founder & CTO of WRDB, a native iOS AI-powered virtual-wardrobe app.
I build the whole stack: a Swift/SwiftUI app in TestFlight beta, a serverless backend on Supabase and Cloudflare Workers, a Next.js web front, and applied AI through computer vision, MCP, and Claude Code.
I build real product, end to end. And I want to keep growing as an engineer.
I'm looking for a curricular internship (mornings) where I can bring real shipping experience to a team.
Let's build something.
Open to a curricular internship: mornings, in Barcelona, through a UB agreement. If you're hiring or just want to talk shop, my inbox is open.