Spoiler: the ingredients are the same.
How much intent, how much taste have you been putting into the things you build?
I’m not asking whether they turned out good. I’m asking something else: how many times did you stop halfway to taste them?
I’ve heard it in a few of the conversation circles I’m part of: Product Engineering is fullstack v2, in a different window. It’s not instant noodles anymore, it’s ramen.
I’m a product engineer at Salvy — a job half the people think is fullstack with a new name. And since I’m curious by default, a year ago I decided to try this eastern delicacy and find out whether they have a point.
Looking at the recipe on the surface, it isn’t hard to conclude the ingredients really are very similar. A pinch of React here, a bit of Node there, SQL with AWS. In short: nothing new under the sun.
Every new challenge brings a set of doubts and feelings with it, and when I took the challenge at Salvy it was no different.
Talking to Danilo at a meetup — someone who would later become my coworker — I noticed there was something there I’d heard about before, but only at much bigger startups: this thing called engineering culture. And even though I’d never tasted it, their concern with the subject made me curious.
At first it was just an aimless conversation about some lousy JS library. I don’t really remember which one.
And that was in fact the moment when, without knowing it, I chose to try. As the bad tongues say, nothing is more powerful than an idea — right?
From then on, the idea of leaving a specialist chair to become a “Product Engineer” — what even is that? — started growing in my head. Before I knew it I was researching where the role came from, whether it was actually new or just another buzzword. Conclusion: nothing new under the sun. A role from 2012, 2013, born at Facebook.
But one thing did make me genuinely curious: the role started making a lot more sense once AI came in to help us.
There was no way around it, I had to try. After a long time settled in frontend, letting myself be wrong and stepping into a new and challenging environment: changing the B2B telco market in Brazil.
The first six months went into understanding the product, the business context, and how I, Celso, could actually deliver value in that operation. Coming from a B2C market, some objections change — and one of them is the discomfort of knowing nothing about the telephony ecosystem. Of having no idea what on earth an MSISDN is. It felt like I had learned technology all over again, and that feeling was very interesting.
Here’s a distillation of one of the reflections I had with my teammates: how much taste are we putting into what we’re doing?
before
I spent my career as a specialist. Building was the hard part: the problem was technical, full of complexity, and it consumed all the energy I had. Take the wrong path and you’d find out three days later — and three days is expensive.
That cost worked as a yardstick I didn’t even notice I was using. You thought before starting because being wrong hurt. You tasted along the way because redoing it was worse. The discipline didn’t come from virtue. It came from the price.
today
Today I generate three versions of an idea in twenty minutes.
The yardstick disappeared along with the cost. And something subtler and more dangerous came with it: technology also removed the need to taste in order to know whether it’s good. You can judge from the outside, objectively, without having gone through the process.
So the question I spent the year trying to answer isn’t whether product engineer is fullstack v2. It’s this one:
If building got cheap, what do we do with what freed up?
None of what I’ve said so far is news. That building got cheap and that judgment became the scarce resource is a conversation a lot of good people already had in 2026, and I agree with it. What I couldn’t find was the next question: where does judgment come from?
a year later
This is the part I didn’t expect, and it’s the reason I wrote this.
It’s not that the day got longer. I still work the same amount, and the queue of things to do hasn’t shrunk — if anything it grew.
What changed is what drains me inside it. That technical, complexity-heavy part, the one that used to take all my energy, stopped taking it. And energy, unlike time, you get to carry home.
That’s when I noticed what I was doing with it. I wasn’t giving it back to the keyboard. I was living more. I spend more time with my family. I have hobbies again — speedcubing, cycling, MPB, a CS degree in the middle of all of it.
And I found out that’s exactly where the palate sharpens. It isn’t rest, it’s input. You’re exercising taste the whole time, on things that have nothing to do with software: choosing, comparing, noticing what works and what only looks like it works.
Going out into the world and trying things became the creative edge. It was the one thing the tooling couldn’t make cheap.
It took me a while to understand why that worked, and the explanation runs through a part of this work we almost never name. When you decide what to build, you’re not executing. You’re creating.
And creating depends on something executing doesn’t: a repertoire.
Creativity doesn’t come from nothing. It’s combination — you put together two things you saw in different places, and the joint is new even if the pieces aren’t. If your repertoire is only what you’ve seen inside the editor, the possible combinations are few. And they’re exactly the same ones available to everybody using the same tools you are.
That’s why the yardstick disappearing was frightening. It’s not that we got worse at building — we got much better. It’s that the part deciding what deserves to be built wasn’t being fed by anything.
A friend sent me this video and the title was already the whole argument. Taste doesn’t come from talent or from studying theory. It comes from consuming a lot and paying attention to what you consumed.
What got me was him talking about technology having removed the need to walk the path. Before, to arrive at a solution you went through the three wrong ones. It was slow and it was expensive — and that was exactly where the palate formed. Not at the destination, on the way.
Whoever generates three versions in twenty minutes covers far more ground and pays attention to far less.
This seems to contradict the thing I like most about speedcubing. Before the timer starts you get fifteen seconds of inspection: you may look, turn it in your hands, plan the entire solve — and you may not touch it to solve. It’s the opposite of going out and trying things.
Except those fifteen seconds only exist because of the thousands of solves that came before. Nobody plans the whole sequence looking at a cube for the first time. Inspection isn’t the opposite of experimenting — it’s what’s left after you’ve experimented a lot.
Looking before you touch is a luxury only repertoire pays for.
what it left behind
This was the first time in my career I found myself taking a risk and stepping out of a specialist position into a more generalist chair. I’ll admit that at first it isn’t easy.
You stop being the person the team calls for the hardest problem in a specific area. You lose your yardstick for progress, because technical depth gave you a measure and breadth doesn’t. And there’s the feeling — legitimate and recurring — that you’re getting shallow at everything.
The honest answer to that fear isn’t “it passes.” It’s that depth is still necessary — it just stops being the axis of your identity. You don’t stop going deep. You choose when to go deep. It was in the same year I became a generalist that I went back to studying operating systems and compilers.
So: it’s the same noodles. And it isn’t the same food.
Ramen broth takes twelve hours, and not because someone stirred for twelve hours. It’s because someone was there. Tasted, waited, tasted again, and at some point decided it was time.
It’s been the period where I’ve learned the most in my career. And I’m not writing that as someone who arrived — I’m writing it as someone in the middle of it.
And I’m writing it because of something I keep noticing, and it bothers me: we consume more and more, and more and more of the same. Same feed, same tools, same references, the same three examples in every conversation. If a palate comes from what you consumed, that should be scarier than it is.
So I’ll hand back the question I haven’t finished answering for myself: where did the taste in the last thing you built come from?
If the answer is “from another product I saw,” that’s fine — a good part of mine comes from there too. Except then your repertoire is the same as everyone else who looked at that same product. And everybody’s broth ends up tasting exactly the same.