JulienSoenen
MSc Corporate Finance, passionate about solving problems, taking ideas from napkin to a startup vision and AI-native product building.
MSc Corporate Finance, UGent '25
My degree gave me the analytical lens: financial modeling, market validation, unit economics. But what I actually do with it is ship. I love to shape visions and ideas into products. Find a real problem, build a first version, validate and iterate. Built two full products that way. Stopped both when the conversations with potential users and market professionals told me the market / technology wasn't there (yet).
Tech-obsessed since day one
I've been solving problems with tech since I was a kid. Jailbreaking iPhones and modding games at 14. That curiosity never stopped, over the years I built a deep understanding of how software, APIs, databases, and AI models actually work together. When AI-native coding tools took off, everything clicked. Now I build real products end-to-end: iOS apps, web platforms, backends - all by leveraging that foundation with tools like Claude Code and Vercel.
What I've built.
Every project started with a real hypothesis and taught me more than any course could. Some are still running, others I stopped after validation.
Bouncer
Validated & StoppedInsurance brokers spend days copy-pasting PDFs into Excel to compare policies. Cognitive hell.
Upload PDFs, get an instant comparison matrix. AI extracts, normalizes, and matches coverage automatically.
15+ broker interviews. 95% don't compare → fixed partnerships with loyalty bonuses. Killed it.
Two-phase AI extraction: first discover the schema, then normalize the data. Handles Dutch insurance jargon automatically. Deployed and functional.
Chaperone
Validated & StoppedRetailers lose €100B/year to theft. Their cameras record it but can't detect it.
CV pipeline that spots product concealment on existing cameras. No new hardware.
Expert validation (consulted Berkeley AI graduate) → false positive rate too high for production retail.
Chained 4 CV models into one pipeline. Built and tested on simulated and real retail CCTV footage.
PlugAlert
ActiveEV drivers who rely on public chargers have no way to know when one frees up. Keep checking manually, or don't charge.
Push notifications the moment a charger frees up. Pick your spots, we watch them.
Real-time PATCH updates via OCPI 2.2 protocol. 90% bandwidth reduction through custom optimization. Launching Scandinavia first → free government charger data.
Sherpa
In DevelopmentVoice memos and meetings turn into walls of text nobody reads, let alone acts on.
AI turns voice into Smart Cards that know what matters for your specific job, with clear CTA's and integrations.
Role Pack system: each profession gets its own extraction schema. 30-second voice memo → structured action card. Full design system and architecture complete. Ready for 2-3 week build sprint.