I help people understand technical products. In my own time, I build them.

Customer Engineer in Basel, Switzerland. Demos, onboarding and technical enablement by day. Everything further down this page I designed, built and shipped myself, outside work, with AI.

View my work LinkedIn

570+documented demos and calls
1,300+technical cases resolved
98%+top CSAT rating, 900+ responses
100k+devices in the largest customer estates

Documented at Patch My PC, 2023 to now.

Flagship project

Wastrel

A daily deck-building card-fight that runs in the browser. Everyone in the world gets the exact same run each day, so the only thing separating two scores is how you played it. No install, no account, a few minutes a go.

What I did

Product direction and the core design decision: one shared daily run, which is what makes scores comparable and gives people a reason to come back
UX and game feel, from onboarding through to the end-of-run share
Art direction for the whole cast and card set
Positioning, analytics instrumentation, and iterating on what the numbers and playtesters actually said

Shipped features daily shared challenge leaderboard and run replays live 1v1 multiplayer

Play Wastrel

Real gameplay

Selected work

Other things I have shipped

Each of these started as a question I wanted answered. They are live, and they are the honest scope of a person building alone with AI, not a team.

Fit or Pass

Tells you honestly whether a job is worth applying for. Two modes: a quick verdict, and an agent.

Why: A single model asked to assess your fit will flatter you. The interesting product problem was making an AI tool willing to say no.
What I did: The evaluation layer is the product, not a feature. A second model audits the first and challenges unsupported claims. The agent mode is a model-driven tool-use loop on Claude that researches, drafts, then checks every line against your evidence before showing it.

Is This A Scam

Paste a message that smells off, get a plain-language verdict, in ten languages.

Why: The people most targeted by scams are usually the last to get a straight answer, and almost never in their own language.
What I did: Scoping it down to one job done well, and the multilingual output, so the answer arrives in the language the victim actually reads.

The claim False

~~$21,000~~

a wallet quietly printing money on Bitcoin markets

What it actually made

$1,125.68 after fees

The strategy, replayed

+$7.850 ms, physically impossible

Two hours of live paper trading, $100 a go, the same decisions filled at six network delays.

0 ms and 40 ms are identical because the order book does not tick in between, so both orders fill against the same snapshot.

Polymarket microstructure research

A viral post claimed a wallet was a bot making $21k on Bitcoin markets. I rebuilt it from public data to check.

Why: The claim was everywhere and nobody had checked the arithmetic.
What I did: Rebuilt the wallet from its public trades, then replayed the strategy forward at six network delays to find out whether the edge was real or just speed. It was speed. Open sourced under MIT so anyone can reproduce it.

Read it open source on GitHub

Layover

A survival game set in the airport of a delayed flight. One thumb, works offline, made for a phone.

Why: I wanted to see whether I could design something genuinely playable in the gap between boarding calls.
What I did: The concept, the constraint that it had to work with no signal, the upgrade loop, and the full cast of original art.

Try one

Neon Run

An endless runner that runs right here, in this page. Jump the red blocks, grab a shield and smash through them instead. Scores go to a global board. It takes about twenty seconds to find out if you are any good.

Space or tap to jump, twice for a double jump. Esc to leave.

Experience

13 years in IT. The last three selling to it.

Most of that was internal work: service desks, network integration, keeping other people's device fleets running. The people I demo to now are the people I used to be, and that is most of why those conversations go well.

Customer Engineer

Patch My PC

2023 to now · Switzerland

Technical pre-sales: inbound discovery, tailored demos, and helping organisations work out whether the product actually fits their environment
Work directly with prospects and existing customers, partnering with Account Managers through evaluations
Translate technical capability into business value, for engineers and executives in the same conversation
Handle technical objections, including the ones where the honest answer is that we are not the right fit
Carry customer and prospect feedback back into the product
Represented the company at Cloud Tech Tallinn, MEM Summit Paris, DTX Manchester and Workplace Ninja Norway

2022 to 2023IT Support SpecialistMDPI
2021 to 2022IT TechnicianInstitute of Molecular and Clinical Ophthalmology Basel
2019 to 2021IT Office ManagerVolkswagen Digital Solutions
2017 to 2018Network Integration EngineerNokia
2015 to 20172nd Level IT SupportSAP
2014 to 2015Customer SuccessPHC Software, SaaS and ERP product demonstrations
2013 to 2014Telecommunications TechnicianNokia Solutions and Networks, Lisbon

About

Portuguese, in Basel.

I am not a trained software engineer, and I do not pretend to be. What I have is ten years inside IT teams and then three in front of the people buying the tools, which turns out to be useful when you go and build something yourself.

I am interested in products, in AI, and in how businesses actually make money. I build things because I like taking an idea the whole way to something real that strangers can use, and because doing it myself is the fastest way to understand what shipping actually costs.

Contact

Say hello.

I like talking about technical products, how they actually get adopted, and what building with AI is really like once you are past the demo. If any of that is your world, my inbox is open.