Jan Aguiló Plana

Jan Aguiló Plana

think different.

Hi, I'm Jan. I don't like convention. I question everything. Combined with my natural ambition and curiosity, this pushes me to “live in the future and build what seems interesting,” as PG puts it.

I am intensely competitive. From friendly football matches to hackathons, I play to win, and I am never satisfied because I always want more.

I believe that who you are drives what you build, not the other way around, which is why I have chosen to present myself by talking about who I am rather than what I do.

The following are the main values/characteristics that define my personality (the order is meaningless). That is, if you ask any person that knows me well enough or that I have worked with to define myself in 3 characteristics, there's a 95% probability that the 3 of them will be included in the list below.

Unconventional follower
Perseverant
Self-confident
Curious
Authentic
Positive / Optimistic
Competitive
Charismatic

Projects

Logistic Integrations & Zendesk App

REVER
REVER Platform

Worked on and took ownership of logistics integrations moving millions in GTV, automating return label creation, tracking, and warehouse inspection flows through robust API connections.

Developed a Zendesk app that pulls live return data from REVER's API into support tickets, showing order details to speed up customer service.

FlowGuard - AB Data Challenge (€6k Winner)

Data Competition
Deep Learning Platform

FlowGuard uses AI to predict water meter failures before they occur, tackling Aigües de Barcelona's core challenge of late anomaly detection. Using unsupervised learning, we analyzed meter readings to estimate risk and identify subtle issues like undercounting, displayed in an interactive map dashboard to help field technicians prioritize interventions.

ZAGUIK: Zero-Knowledge Attestation for Privacy-Preserving LLM Governance

Research Paper
Verifiable AI

Zaguik is a privacy-preserving attestation system that uses zero-knowledge proofs to cryptographically prove that an LLM output passed a PII filter, without revealing the output itself or the filter's internals, enabling verifiable AI governance.

Consumer behaviour on LLMs - AI Search Analytics Hackathon (€10k winner)

Hackathon
AI Analytics Platform

Built a working Chrome extension that shows discount coupons for products featured in AI Search, capturing and analyzing user prompts in the process. Classified user intent (consideration, evaluation, decision) and clustered prompts to generate actionable brand insights and content recommendations.

Blog & Thoughts

Here I share experiences that shape me, along with the lessons and reflections behind them, or simply some of the unusual thoughts that cross my mind. Some are in Catalan, others in English, depending on how I feel when I write them. All are written without AI, so they capture exactly what I am thinking in that moment without bias. I don't ask you to like them, I just want you to get inspired by them.

Loading reflections...

Contact