The pitch was one sentence, which is the only kind worth building: connect a database, get an API. Migrations handled. Documentation generated. No backend to write.
I was a founding engineer. That title means the work has no shape yet, and neither does your job description.
The first UI
I designed it in Figma and built it in Next.js.
I am not a designer. I mention this because founding engineers are frequently asked to be one for a while, and the useful skill is not taste — it is knowing that a mediocre interface you can iterate on beats a beautiful one you are waiting for. We shipped something plain and legible, and then we watched people fail to use it, and then we fixed the places where they failed.
Rust, and where the domain boundaries go
The engine — custom queries, migrations, the machinery that turned a schema into an API — was Rust.
Rust is opinionated about ownership, and a codebase that is opinionated about ownership will force you to be honest about your domain boundaries much earlier than a TypeScript one will. You cannot casually pass a mutable reference to a thing that three modules believe they own. The compiler makes you decide who owns it. I have written more forgiving code in more forgiving languages and it was worse code.
TODO — Alberto: The custom queries project is the best technical story you have here. You were building raw SQL from Handlebars templates, hit a hydration bug that produced syntactically invalid queries, and fixed it upstream — that PR is on this site. Tell that story properly: what were you templating, why did the hydration break the SQL, what did the patch change? It shows a specific, unglamorous kind of competence: you hit a wall in someone else's library, read it, fixed it, and got it merged rather than working around it.
Hiring
I hired our founding product team — engineers and a designer.
TODO — Alberto: What did you learn about hiring that you did not know before? What did you screen for that turned out not to matter? Two paragraphs. Most engineers cannot write this section at all, which is exactly why it is worth writing.
It shut down
Neurelo shut down in 2024.
I could leave this off the site. Most people would. But a portfolio that only contains successes is not a record of a career, it is a marketing document, and everyone reading it knows the difference.
We built the thing. It worked. The market did not arrive, or it arrived somewhere else, or we were early — I have my theories and none of them are load-bearing. What I took from it is that the quality of the engineering and the survival of the company are less correlated than engineers would like, and that this is not an argument for engineering less carefully. It is an argument for choosing carefully what you point the care at.
TODO — Alberto: Replace the theorizing above with what you actually believe happened. Be specific and be generous to the people involved. This is the section a good engineering manager will read most closely, because it is the only one that tells them how you think about failure.