I work on problems where technical decisions have economic consequences, and where one person who understands both can move faster than a team split across functions.

My background is in engineering and industrial systems. Started as an apprentice at Novelis in Sierre (Valais, prettiest canton of all). Then at SFC KOENIG (Zürich), I owned products from concept to scale: defining requirements, running predictive simulations to prevent 20,000+ annual failures, building data tools from raw production statistics, launching product lines into Chinese markets at 2M+ parts per year in compliance with strict IATF norm requirements. Not as an engineer waiting for specifications, but as the person writing them and shipping the result.

That experience shaped how I think: in constraints, trade-offs, and responsibility. The bottleneck is rarely technical. It is deciding what to build, why it matters, and taking ownership of the outcome.

During my master's, I treated the “open curriculum” as an optimization problem. I worked 60–80% in parallel with my studies to gain real exposure, and used academic freedom to go as deep as possible into computational systems: optimization, predictive modeling, machine learning, and market forecasting. My interest in computation started earlier with fluid mechanics and FEA, and has since evolved into large-scale ML systems and energy-aware architectures.

Finishing an MSc in Business Engineering at ZHAW (2026). Speaking French, English, German and Portuguese. Based in Zürich. Besides my friends, my family and the fresh air, I love running and reading. Also, I dig cinema.

Today, I am at NTUST's LLM Lab in Taipei, working on energy-efficient inference for retrieval-augmented generation systems. The question I am chasing: can we replace static pipelines with conditional architectures that adapt computation to the query? The goal is simple: reduce energy per query by 60% or more while preserving answer quality.