Part 1Marc de Batlle
0%
Part 1Marc de Batlle
One

Data-driven business,built end to end,shipped to users.

Wikidata mark
Chapter concepts
  1. 1Fourth-year BSc in Business Intelligence & Data Analytics at La Salle BCN, expected July 2027.
  2. 2Hybrid profile: understand the business problem, translate it into a technical solution, ship it.
  3. 3Freelance developer for real clients: a CRM for a 250-client tax advisory; a speech platform and a members' area spec for Federació ACAPPS.
  4. 4Co-founder and CEO of WRDB, an AI-powered iOS app live on TestFlight; I own data, ML, analytics and growth.
  5. 5SQL, Python, R, Swift, TypeScript; Power BI, Tableau, Power Platform; Supabase, Next.js, Docker; ML, computer vision, LLM tooling.
  6. 6Seeking an internship in data or AI under La Salle's university-company agreement.

Fourth-year Business Intelligence & Data Analytics student with a hybrid profile: I understand the business problem, translate it into a technical solution, and ship it. Not the analyst who hands over a slide and walks away, and not the developer who never asks why. Freelance, I have shipped a CRM that ran a full tax season for a 250-client advisory, a speech platform collecting non-normative Catalan for the BSC, and the specification of a CRM-integrated members' area. At WRDB, an AI wardrobe app on TestFlight, I am co-founder and CEO and own data, ML, analytics and growth. I built BuildLoud, a marketing product, in 48 hours at Basejump, a hackathon backed by V3V Ventures and Lost Astronaut, and volunteered on Fundació Comtal's Sant Jordi campaign. I am seeking an internship in data or AI.

Experience

Newest first. Boxes open to a full case study: the data, the pipeline, and the numbers that came out the other end.

2026 to present

Freelance Developer

Business & AI Solutions

Real clients, real deadlines. I take a business problem, translate it into a technical solution on whatever stack the client already pays for, and ship it. Two clients so far, three deliverables, each with its own case study.

Renda 2025 CRM

Consultoria Sostenible, a one-person tax advisory

Autónomos rarely get a seat at the enterprise CRM table, so I built it the other way around: a campaign tracker on the Microsoft 365 tools the advisory already paid for. Live dashboard, email tracking with overdue filters, one-click invoicing, 250+ clients through tax season, zero new subscriptions. The dashboard is a filterable gallery over the SharePoint list, coloured by eleven campaign states, with a priority queue of who needs attention today. One click fires a flow that fills the Excel template, files it in the campaign folder and marks the client done. The gain is focus: instead of hunting through folders and an inbox for each client's file, she reads one screen and gets on with the returns.

Consultoria Sostenible

Scale
250+ clients, one tax season
Time saved
about 40 h a season on invoicing and follow-up
Tools
Power Apps, Power Automate, SharePoint

Federació ACAPPS

Two products for the Catalan federation for people with hearing loss

Donants de Veu: I built veus.acapps.org, a self-hosted Mozilla Common Voice rebuilt to collect non-normative Catalan speech for the Barcelona Supercomputing Center, with a phase-based sentence swap the platform does not support natively, a frontend to the client's brief and an anonymised WAV export with a manifest and a private mapping, every file verified against its source. ARA Cultura: the technical specification of a members' area integrated with SinergiaCRM, replacing a manual process for group bookings of accessible events. Five CRM entities, nine screens and seven automations, written so the in-house developer could build from it.

Deliverables for Federació ACAPPS*2
(Donants de Veu for the BSC; ARA Cultura)
Speakers in the corpus, eight groups80
StatusDelivered, 2026

*Docker, MySQL, React, Python and ffmpeg; ARA Cultura specified on SinergiaCRM. Partner: Barcelona Supercomputing Center.

Clients: Consultoria Sostenible; ACAPPS.

01
Aug 2025 to present

Co-founder & CEO, WRDB

AI virtual wardrobe for iOS, live in beta on TestFlight

WRDB photographs your clothes, recognizes them, and recommends outfits from what you actually own. Three founders: I am CEO and CFO, and I own data, machine learning and growth. I wrote the first version of the app in the summer of 2025 and have shipped 745 commits since, across Swift, SQL, TypeScript and Python.

Analytics and the growth machine

TelemetryDeck, Supabase, Next.js, Meta and TikTok APIs

Instrumented the app with 33 analytics signals, built the cron warehouse for DAU, retention, churn and activation, and a 32-page dashboard with a live activity feed on Supabase Realtime and Slack alerts. The growth side is a six-layer marketing-machine: scheduled publishing to Instagram, Facebook and TikTok, analytics pulls, anomaly detection, and a daily auto-generator that renders reels for an AI persona. Alerts go to Slack rather than living in the dashboard, and the whole analytics layer shipped in one day across fourteen phases.

Analytics signals, consent-gated*33
Dashboard pages on Next.js32
(one of them realtime, on Supabase)
Marketing edge functions on pg_cron8
StatusLive, internal

*TelemetryDeck into a Supabase cron warehouse; built with Next.js, Recharts and SQL.

ML and the app

Swift, Core ML, Supabase

Built the original app and its Clean Architecture spine (291 Swift files, the DI container, the domain layer). Converted FashionCLIP from PyTorch to Core ML so garment embeddings run on the phone, and wrote the outfit generator: weather from WeatherKit, occasion, color and style scoring, and a preference learner that re-ranks from what the user accepts and wears. The conversion pipeline wraps the model, exports an fp16 mlpackage and checks parity against PyTorch, and the model unloads after a minute idle to save memory. The affiliate marketplace is mine too: the Awin partnerships, a curated catalogue with CLIP vectors, and the recommendation engine that blends discovery with the user's own taste.

FashionCLIP, on the phone

Embeddings
512-d CLIP, on-device
Scored
up to 500 combinations per session
Tools
Swift, SwiftUI, Core ML, PyTorch
RoleCEO & CFO, co-founder
OwnsData, ML, analytics, growth
StatusTestFlight beta
Swift·Core ML·Supabase·Next.js·TypeScript·SQL

See wrdb.site and the TestFlight beta.

02
Volunteering
Jan to Apr 2026, Fundació Comtal

Sant Jordi 2026 campaign volunteer

Handmade fabric roses sold to companies to raise funds for people in vulnerable situations.

Part-time work
Apr to Sep 2024, Jan to Apr 2026, La Pallaresa, Sant Lluís SA

Waiter

Two seasons on the floor. Service under pressure, the same skill that ships a product in 48 hours.

International
Jun 2025, UESTC, Chengdu, China

Machine Learning & Business Intelligence course

International program at the University of Electronic Science and Technology of China.

Experience
Oct 2024 to Feb 2025

Co-founder, DHM Middleman

Client acquisition and sales for small businesses: cold calling, sales funnels and delivery of simple, high-impact solutions.

2023 to Jul 2027 (expected)

BSc in Business Intelligence & Data Analytics

La Salle BCN

Statistics, machine learning, data engineering and the business side of turning numbers into decisions. Most of what I learn ends up in a project the same week.

Honors: 1st place, E-Business course, with MEALPRINT, an AI app concept that scans meals and estimates their carbon footprint in real time.

Coursework highlight: logistics carrier analysis (1P vs 3P) across roughly 525K packages, with Tableau dashboards and an executive report.

DegreeBSc BI & Data Analytics
SchoolLa Salle BCN
ExpectedJul 2027
Tableau·Power BI·Python·SQL·R
03

Projects

Self-directed builds. More repositories on GitHub under wyndollin.

Apr 2026
Exhibit 04

BuildLoud

The WRDB marketing machine, rebuilt in 48 hours as a product anyone can switch on.

At Basejump Hacker House Barcelona, backed by V3V Ventures and Lost Astronaut, we forked the marketing machine that runs WRDB's Instagram and turned it into a multi-tenant product over one weekend: describe your brand, point it at your GitHub repo, connect Instagram, and a face-locked week of content plans, renders, publishes and learns on its own. Live at buildloud.site since the Saturday morning.

48 h
Friday evening to Sunday afternoon
283
commits on the Saturday
24
two-hour sprints, all closed
2025 · May
08

Trip-Setter

A web app that personalizes your trip.

A Python web app that takes your preferences and builds a trip around them: a small personal project that solved a real need and kept my web skills in use between data builds.

StatusShipped
RoleSolo build
Open case study 08

About

Fourth-year Business Intelligence and Data Analytics student at La Salle BCN with a hybrid business and technical profile. My work usually starts with a number that looks wrong: a market that overreacts, a wardrobe nobody uses, a tax office drowning in spreadsheets. It ends with a model, a scraper, or an app that does something about it.

That is what I do as a freelance developer for real clients and as co-founder and CEO of WRDB, an AI-powered iOS app live in beta. I am seeking an internship in data or AI under La Salle's university-company agreement. Spanish and Catalan are native; English is professional.

BasedBarcelona, ES
StudyingBSc BI & Data Analytics, La Salle BCN
GraduatingJul 2027 (expected)
NowFreelance developer, CEO of WRDB
SeekingInternship in data or AI
LanguagesSpanish, Catalan (native), English (professional)

The stack

Solve for x = a shipped model. Given: the tools below, grouped by the part of the problem each one is used on.

Languages
  1. 1.SQLPostgres, MySQL, SharePoint lists
  2. 2.Pythonpandas, scripts, ML
  3. 3.RStatistics
  4. 4.Swift / SwiftUIWRDB app
  5. 5.TypeScriptNext.js, edge functions
BI & Analytics
  1. 1.Power BIDashboards
  2. 2.Tableau525K-package analysis
  3. 3.ExcelModels, templates
  4. 4.PandasPreprocessing
  5. 5.Predictive modelingClassifiers, clustering
  6. 6.TelemetryDeckProduct analytics
Power Platform
  1. 1.Power AppsCanvas apps
  2. 2.Power AutomateFlows, Office Scripts
  3. 3.SharePointData source
Web & Data Eng.
  1. 1.SupabasePostgres, auth, cron
  2. 2.Next.js / ReactDashboards, sites
  3. 3.Docker ComposeCommon Voice
  4. 4.MySQLCorpus, export
  5. 5.Web scrapingrequests, BeautifulSoup
Applied AI
  1. 1.Machine learningEnsembles, embeddings
  2. 2.Computer visionGarment analysis
  3. 3.Core MLPyTorch to on-device
  4. 4.Claude CodeLLM tooling, MCP

*Answer is given in shipped software, rather than % of par value.

Summary

Seeking an internship in data or AI under La Salle's university-company agreement. Roles, collaborations, or just a conversation about data, machine learning or shipping fast: my inbox is open and I reply. If the work above matches what you are looking for, the routes below are all current.