Portfolio '26

ThibaultVanHellemont

Software · automation · AIScroll

I see where things can be better, and then I build it.

Developer & IT generalist : software, automation and AI, end to end.

Selected work

03 Work

SEO audit platform · marketing agency

Stack

PythonFastAPIReactClaudeSEMrushPlaywrightSQLiteDocker

From domain to shareable SEO report in five minutes

2026 · Built solo · from first idea to production

Enter a domain and five minutes later you get a full, shareable SEO and marketing report. Work a consultant used to do by hand for hours.

From domain to shareable SEO report in five minutes, screenshot 1

From domain to shareable SEO report in five minutes, screenshot 2

From domain to shareable SEO report in five minutes, screenshot 3

From domain to shareable SEO report in five minutes, screenshot 4

From domain to shareable SEO report in five minutes, screenshot 5

From domain to shareable SEO report in five minutes, screenshot 6

From domain to shareable SEO report in five minutes, screenshot 7

From domain to shareable SEO report in five minutes, screenshot 8

From domain to shareable SEO report in five minutes, screenshot 9

01 / 09

A handful of screens from one real audit, from dashboard and intake to the full report. The actual report is interactive, multilingual and shareable. These are just a few frames to show the flow.

The problem

A full SEO and marketing analysis cost a consultant hours of manual work: checking tech, gathering keywords, comparing competitors, pouring it all into a report.

The solution

Enter a domain and the tool measures the full technical and strategic SEO, runs an AI analysis with Claude (keywords, content, competitor gaps), scores five pillars and serves the result as an interactive, multilingual, shareable report with PDF export.

Under the hood

FastAPI backend with clean, testable architecture (domain logic separate from I/O)

AI analysis via Claude with data sanitisation & prompt-injection protection

Parallel scraping with JS rendering (Playwright) and SSRF protection

Integrations: SEMrush, PageSpeed, OpenPageRank, with caching & cost control

Scoring engine with configurable weights across five pillars

React frontend with live WebSocket updates, multilingual report (NL/FR/EN)

Full deployment in Docker, with user management and logging

How it works

Enter a domain

Drop in a website, pick the target region, report language and optional competitors. The rest runs on its own.

Measure & scrape

A parallel crawl with JS rendering measures tech, speed, content and backlinks, enriched with SEMrush, PageSpeed and OpenPageRank data.

AI analysis

Claude reads the raw measurements and writes the strategic analysis: keywords, content gaps and competitor comparison, in the right language.

Score

A scoring engine computes scores across five pillars, with weights the consultant can tune per assignment.

Report

The result becomes an interactive, multilingual, shareable report with PDF export, ready for the client.

Marketing attribution dashboard · Time Out Box

Stack

SupabaseReactGoogleMetacron

The entire funnel in a single dashboard

2026 · Built it and run the channels myself (SEO, GEO, Google & Meta Ads)

One dashboard that merges four marketing sources and shows the whole funnel, from impression to sale. Refreshed automatically instead of by hand each week.

The entire funnel in a single dashboard, screenshot 1

The entire funnel in a single dashboard, screenshot 2

The entire funnel in a single dashboard, screenshot 3

The entire funnel in a single dashboard, screenshot 4

01 / 04

The problem

The full funnel was scattered across four platforms. Gathering numbers for a report every week ate time and made fast course-correction impossible.

The solution

One private dashboard showing the whole funnel: from impression to click to visit to contact to sale. Data from GA4, Google Ads, Meta Ads and Search Console is fetched and merged automatically, with funnel attribution, ROI per channel, an alert system, an interactive map of Belgium with live visitors and monthly goals with projection.

Under the hood

Automatic data pipelines (cron) from four sources, stored in Supabase

Funnel attribution linking leads to their source (impression → sale)

Cost & ROI analysis per channel, with period comparison and trends

Alert system (no spend, cost per contact too high, traffic dropped)

Interactive map of Belgium with sales and live visitors

Monthly goals with progress and projection, secured access via allowlist

Internship & thesis · Made (Rombit Studio)

Stack

TypeScriptReactSpring BootPostgreSQLFirebase

Live score and Elo rating for the office foosball table

2024 · 14-week internship · full stack, Agile, in a real dev team

A full-stack app that turns the office foosball matches into a live score and a fair Elo ranking. My first project inside a real dev team.

Live score and Elo rating for the office foosball table, screen 1

Live score and Elo rating for the office foosball table, screen 2

Live score and Elo rating for the office foosball table, screen 3

01 / 03

The problem

Made's office played fierce post-lunch foosball, but had no way to track matches, scores or who was actually the best. Who really is number one?

The solution

A full-stack PWA for the 'Rombit Kicker Competition': players log matches with a live score, the backend keeps an Elo rating (with full history) and turns it into a fair, real-time ranking. Goals even arrive automatically from an Arduino detector over a message queue, and a push notification fires when someone is about to crawl under the table. My first time in a professional dev team: I worked Agile in two-week sprints, picked up issues in Jira and estimated them with story points, with a code review on everything I shipped so my code stayed clean and solid.

Under the hood

React + TypeScript PWA, installable, talking to the backend over WebSockets

Java Spring Boot backend with a PostgreSQL database

Elo rating per player with full history, feeding a live ranking

Match flow with live score updates and head-to-head stats

Push notifications via Firebase (e.g. when someone's about to crawl under the table)

Automatic goal detection: Arduino/IoT messages processed off a queue

NFC login and reservations

Agile in two-week sprints: issues and story points in Jira, with a code review on every change

Bachelor's thesis (PDF)

About me

What I enjoy most is building.

My foundation is technical, and has been for years: web and app development, server setup, networking, data integration. On top of that came a background in media and marketing, and that's where it got interesting. Because I can't just build something, I also see what's worth building and why.

I use that combination to spot where processes can be smarter, and then build the solution myself, from first idea to a working system in production. Most of my work sits in software, systems and automation. AI is the recent, powerful layer on top. I'm as happy talking to developers as to marketers.

I mostly learn new things by building with them. A framework, an API, a piece of hardware: I only really get it once I've made something that works. That curiosity keeps pulling me toward the next project, and I try to automate just about anything that can be automated.

Born 2002 · automates just about everything

Skills

What I use

Development

HTML · CSS · JavaScript · TypeScript

React / Vite · Next.js

Python / FastAPI

App development

Supabase · SQLite

Netwerken · server-setup · Docker

AI & automation

LLM-API's (Claude)

Procesautomatisatie

Data-pipelines & cron-syncs

Bouwen met AI-tools (Claude Code)

Web & content

Webflow

Wix

WordPress

UX-gericht digitaal denken

Marketing & data

SEO · GEO

Google Ads · Meta Ads

GA4 · Search Console · SEMrush

Adobe & Canva

Languages

Dutch

native

French

native

English

proficient