Sign in

João Alves

@joaoqalves.net
72 followers 27 following 126 posts

2x Dad. Husband. Head of Engineering @Adevinta. Building abistama.com and rotahog.com Opinions are my own. Writes in PT/ES/EN/CAT

PostsRepliesMedia
João Alves @joaoqalves.net · 18/02/2026
Most change in a 200-person org needs to pass through 4 layers before it reaches teams. By the time AI adoption trickles down, the window may already be closing. This isn’t something you delegate. See what @tobi is doing. You have to lead it. 👇open.substack.com/pub/joaoqal...
000
João Alves @joaoqalves.net · 18/02/2026
Should engineering managers become hands-on again? If even 20% of what we’re seeing is real, productivity, speed, and cost structures will change. Leadership blind spots at inflection points are expensive. I think eng leaders can’t afford to find it out.
110
João Alves @joaoqalves.net · 28/01/2026
So I wrote about software in a post-abundance world: why cheap code reshapes teams, labor economics, where capital flows next, and why the disruption won’t come from “AI features” alone. Wanna read? newsletter.terminalprompt.com/p/software-...
newsletter.terminalprompt.com
Software in a post-abundance world
Why AI changes teams, labor, and capital at the same time
000
João Alves @joaoqalves.net · 28/01/2026
🧵 Yesterday, after a discussion on Discord, and today with @sdepablos, I got inspired. AI is not just changing how software is built. It’s changing how the industry operated for years. The loop is: smaller teams → cheaper feature parity → margin pressure → labor shifts → capital reallocates. 👇
110
João Alves @joaoqalves.net · 27/01/2026
After reading the article on software industrialization, this is my (current) take on Software Engineering + AI. Where am I wrong?
000
João Alves @joaoqalves.net · 20/01/2026
Full post here: open.substack.com/pub/joaoqal...
000
João Alves @joaoqalves.net · 20/01/2026
“I already raised my concerns. If they chose to ignore them, that’s on them.” I heard it many times from Tech Leads. But this is not (tech) leadership at all. So... I wrote about it. 👇
110
João Alves @joaoqalves.net · 13/01/2026
Wait... what is "operational excellence"? It is about: * Defining what “good” means * Knowing when to push * Knowing when to stop and fix things SLOs and error budgets make trade-offs visible to Tech Leads. Full article: open.substack.com/pub/joaoqal...
000
João Alves @joaoqalves.net · 13/01/2026
As a Tech Lead, “we need more speed” always comes up. Without KPIs, SLIs, and SLOs, that turns into opinions: “I think it’s fine.” // “I think it’s risky.” This is why Operational Excellence matters. 👇
110
João Alves @joaoqalves.net · 09/01/2026
ProTip™: comfort is not a strategy. This is "war-time" for CTOs. Full post: newsletter.terminalprompt.com/p/ai-wont-b...
newsletter.terminalprompt.com
AI won't break your company, but pretending nothing changed will
How CTOs should rethink teams, delivery, and leadership in an AI-native world
000
João Alves @joaoqalves.net · 09/01/2026
AI won’t break your company (yet?). It will expose slow org design, fragile delivery pipelines, and leadership habits that no longer scale. I wrote down what actually needs to change for CTOs running 100+ engineers. 👇
120
João Alves @joaoqalves.net · 18/12/2025
Excelencia Operacional desde la lente del Tech Lead: * definir qué significa “bueno” * medirlo con señales útiles * convertirlo en decisiones repetibles Artículo completo: enespanol.joaoqalves.net/p/excelenci...
enespanol.joaoqalves.net
Excelencia Operacional, la palanca del Tech Lead
Cómo usar KPIs, SLIs y SLOs para convertir métricas en decisiones
000
João Alves @joaoqalves.net · 18/12/2025
Si eres Tech Lead y tus métricas no cambian las decisiones, no te están ayudando. KPIs, SLIs y SLOs no van de medir más. Van de dejar de discutir opiniones y empezar a operar el sistema. Nuevo artículo 👇
100
João Alves @joaoqalves.net · 16/12/2025
🎁 I also prepared a small gift for anyone who wants to put this into practice right away. A practical plan for your first 90 days as a Tech Lead, together with my EM/TL alignment toolkit. Read it here -> world.hey.com/joaoqalves/...
world.hey.com
I’m a Tech Lead, and nobody listens to me. What should I do?
In June 2018, I joined mytaxi (FREE NOW), a competitor of Uber in the ride-hailing space, as Backend Chapter Lead. I was looking for an opportunity to grow in technical leadership. Honestly, I did not even fully understand what “Chapter Lead” meant. After some research, I learned it was part of Spotify’s squad (team) and chapter (horiz...
010
João Alves @joaoqalves.net · 16/12/2025
I joined mytaxi as a Tech Lead in 2018 and quickly realized something uncomfortable: having good ideas and the title did not mean people would listen. What finally changed things was simpler. A small tool and sitting side by side with people, solving problems. Full article 👇
130
João Alves @joaoqalves.net · 15/12/2025
“Why do you blog?” I learned most of what I know by watching great teams up close and trying to apply their habits in the real world. Writing is my way of passing that learning forward. Happy to see a positive impact.
010
João Alves @joaoqalves.net · 11/12/2025
🎁 Además, regalo un plan práctico de 90 días para cualquier Tech Lead que quiera empezar a tener impacto mañana mismo. open.substack.com/pub/esjoaoqa...
open.substack.com
Soy Tech Lead y no me hacen caso. ¿Qué hago?
Cómo ganarte el derecho a influir en tu equipo
000
João Alves @joaoqalves.net · 11/12/2025
El otro día, un pequeño post de Félix López me inspiró. Los títulos no te dan influencia por arte de magia. Si eres TL y tu equipo no te hace caso, no suele ser porque tu estrategia sea mala. Suele ser porque aún no existe confianza. ¿Qué me gustaría haber sabido en 2018? 👇
100
João Alves @joaoqalves.net · 09/12/2025
It’s a translation of my post in Spanish, and it also includes a short FAQ: * What changes when EM = TL? * How to avoid “imposing” decisions? * How to set principles that create autonomy? Full post here → world.hey.com/joaoqalves/...
world.hey.com
Traits of a good Tech Lead
The software development industry during the 2010–2020 decade created roles such as Engineering Manager (EM) and Tech Lead (TL). People had already been using one or the other, but during that decade, most companies adopted them. While they vary by company, they share common elements. Note: this article is a translation from the origin...
010
João Alves @joaoqalves.net · 09/12/2025
The Tech Lead role is among the most misunderstood in engineering teams. It’s not “the best coder”, and it should never be a bottleneck. I wrote a post on what good Tech Leads actually do, the signals to look for, and how EMs can better support them.
110
João Alves @joaoqalves.net · 04/12/2025
Al final, ofrezco un “toolkit” de alineación entre EM y Tech Lead, con un semáforo de señales, principios operativos y ejemplos prácticos para evaluar la autonomía, la arquitectura, la calidad y el scope técnico. 👉 open.substack.com/pub/esjoaoq...
000
João Alves @joaoqalves.net · 04/12/2025
¿Qué hace realmente un Tech Lead? Y, sobre todo, ¿cómo saber si lo estás haciendo bien? Hablo de arquitectura, calidad, alcance, principios operativos y del verdadero impacto de un TL: crear un equipo que toma mejores decisiones sin depender de él. 👇
100
João Alves @joaoqalves.net · 02/12/2025
2/2 I love the power-law analogy :) How to navigate this shift in the industry? I wrote about it here: world.hey.com/joaoqalves/...
world.hey.com
When software becomes fast food
Generative artificial intelligence has amazed the World. Since OpenAI launched ChatGPT 3, its user adoption has been staggering. In 2022, ChatGPT surpassed one million users in just five days. For comparison, Instagram needed 2.5 months back in 2010. And it’s not just OpenAI. Anthropic has Claude, one of the best models for programming...
000
João Alves @joaoqalves.net · 02/12/2025
1/2 Considering the success of my article in Spanish, I’ve translated it to English: “When software becomes fast food” AI commoditizes code. So the bottleneck shifts and intensifies elsewhere (CI/CD, architecture, taste, judgment, etc.).
111
João Alves @joaoqalves.net · 27/11/2025
Hoy os traigo un artículo sobre algo que a veces olvidamos: no debemos enamorarnos de la tecnología, sino de nuestros usuarios. Suena fácil y obvio, pero los equipos de ingeniería se olvidan a menudo de ello. open.substack.com/pub/esjoaoqa...
open.substack.com
Bricks of Love: enamórate de tus usuarios, no de tu tecnología
Cómo las actualizaciones semanales pueden reforzar la cultura, el propósito y la empatía en tu equipo
010
João Alves @joaoqalves.net · 25/11/2025
The ladder didn’t disappear. It just got steeper. If you’re an EM or staff IC aiming higher, I’d love to hear your stories as well. world.hey.com/joaoqalves/...
world.hey.com
Lessons from two failed promotions... and What Changed After ZIRP
Every time we talk about promotions and career paths, we listen a lot about how to get there. Today, I want to tell the story of how I failed to get the role... twice! For context: I joined Adevinta as Engineering Manager of the Runtime team in late 2020. We were part of an internal Platform-as-a-Service (PaaS) offering. We provided SC...
000
João Alves @joaoqalves.net · 25/11/2025
I failed two promotion attempts even when everything “looked right” on paper. In a post-ZIRP world, doing great work isn’t enough. You need level n+1 stories, visibility outside your team, and impact across boundaries. Full article below.
Whiteboard sketch outlining three promotion mistakes: "potential" trap, competition favors proven experience, and visibility gap limiting cross-team trust.
110
João Alves @joaoqalves.net · 24/11/2025
Nueva edición de “El doble clic” - Gestión de contexto en Claude Code, cómo Tinybird reforzó su infraestructura en entornos multi-tenant y mucho más. Aquí 👇 enespanol.joaoqalves.net/p/el-doble-...
010
João Alves @joaoqalves.net · 20/11/2025
Private Equity en tech: todo lo que siempre quisiste saber enespanol.joaoqalves.net/p/private-eq...
enespanol.joaoqalves.net
Private Equity en tech: todo lo que siempre quisiste saber
Qué es, cómo funciona y qué implica para quienes trabajamos en tecnología
010
João Alves @joaoqalves.net · 19/11/2025
Doing active/active K8s setup is hard. In many cases — unless it’s a regional service with its own database, etc — it comes with a lot of hidden complexities. @tjamet writes about our experience at Adevinta doing it. Must read: medium.com/learnings-f...
110
João Alves @joaoqalves.net · 18/11/2025
Cloudflare down creating this huge impact is a nice reminder of why I’m a happy shareholder since a long time ago.
010
João Alves @joaoqalves.net · 18/11/2025
“Meta evaluating AI fluency” is making people mad about micromanagement. Using metrics as prompts, not targets, can solve issues and boost performance. Like DORA metrics at Adevinta, AI fluency needs clarity. Full article: world.hey.com/joaoqalves/...
world.hey.com
AI adoption needs light, not hope
As per last week’s rumors, Meta will judge employee performance based on AI skills. Nothing new. Other companies, like Zapier, have even published what they expect “AI fluency” for all new hires: Every time a company announces “AI usage dashboards,” people get mad. My timeline on X gets full of surveillance, micromanagement, and bullsh...
110
João Alves @joaoqalves.net · 17/11/2025
Managing EN/ES accounts and blogs with diverse content is more challenging than I anticipated, but I’m really enjoying it. Tomorrow, I'll post about how Meta and others are measuring AI adoption and making it a requirement in performance reviews. Stay tuned!
010
João Alves @joaoqalves.net · 17/11/2025
A tí por la charla! Muy buena
010
João Alves @joaoqalves.net · 17/11/2025
Hoy lanzo “El doble clic” en mi blog. Cada semana compartiré lecturas, charlas y enlaces que me hayan gustado. Perspectivas de @paoramen.com, @eferro.net, @elwatto.bsky.social y más! Si te gustan la tecnología, los productos y las startups, ¡te interesará! 👇
enespanol.joaoqalves.net
El doble clic — Semana del 17 de noviembre
Ideas, artículos y descubrimientos de la semana
231
João Alves @joaoqalves.net · 14/11/2025
Mi tesis sobre la IA: abarata picar código y desplaza el valor hacia el criterio. Habrá mucho “software fast-food”, pero también una mayor necesidad de expertos que sepan diseñar, integrar y tomar decisiones. open.substack.com/pub/esjoaoqa...
open.substack.com
Cuando el software se vuelve “fast food”
La nueva frontera no es programar, sino decidir qué merece construirse
000
João Alves @joaoqalves.net · 09/11/2025
More updated than ever. 2020s engineering leaders must be more into the details. That’s a feature and not a bug. “Servant leadership” (misapplied) created a lot of peace time engineering leaders. Full post on world.hey.com/joaoqalves/...
120
João Alves @joaoqalves.net · 03/11/2025
On the AI Agents vs workflows, my mental model is a sports team. * An agent is like a player knowing specific skills (speed, dribbling, …) * A workflow is like a tactic, using AI agents to achieve a goal (eg, score)
000
João Alves @joaoqalves.net · 02/11/2025
Cloud/FinOps Pro Tip: run Claud Code on AWS Bedrock, as they have the models available. Two benefits: 1. Built-in access control with AWS and temporary credentials. 2. It contributes to your AWS expenditure/commitment, allowing you to leverage this cost for volume discounts.
010
João Alves @joaoqalves.net · 02/11/2025
Un poco a contracorriente: no todo el esfuerzo es tóxico. open.substack.com/pub/esjoaoqa...
open.substack.com
El esfuerzo no es tóxico
Nota: Este artículo es una traducción al español del original en inglés, escrito por mí.
000
João Alves @joaoqalves.net · 31/10/2025
One sad thing is that I’m Portuguese, and we tend to write long sentences, list things, use a lot of adjectives… apparently this is also a trait of ChatGPT somehow 😢 Not nice.
000
João Alves @joaoqalves.net · 31/10/2025
Usually, my flow is: 1. Something happens at work or a friend tells me a story about theirs and we discuss 2. I think about that thing for 1 week to 3 months. Mostly when I’m taking a shower 😅 3. Then I can write an article in less than one hour.
100
João Alves @joaoqalves.net · 31/10/2025
“How do you write so many articles in such a short time frame? You must be using AI” Nope. I use Grammarly to (re-)write as I’m not a native English guy. I use ChatGPT mainly to create an illustration, as otherwise HEY World (my blogging provider) puts my photo on it.
100
João Alves @joaoqalves.net · 29/10/2025
The sequel is out. Four new "disasters” I've seen first-hand: #7 more services than engineers #8 the gateway to hell #9 technology sprawl #10 when the org chart becomes your architecture Full post: world.hey.com/joaoqalves/...
world.hey.com
Disasters I've seen in a microservices world, part II
When I first wrote about microservice disasters, I thought we'd eventually "solve" them, with better tooling, frameworks, and operational maturity. We didn't. We just learned to live with the chaos. Distributed systems will always surprise you: timeouts, retries, and fallacies don't disappear; they just shift shape. Maybe that's the re...
000
João Alves @joaoqalves.net · 29/10/2025
"538 microservices? What a mess." Four years ago, I wrote Disasters I've Seen in a Microservices World. I thought by now we'd have solved most of them. We didn't. We just learned to live with the chaos. Link below 👇
A graph showing how risk changes: too few services? monolith, and potentially problems when deploying. A lot of services? A headache too
111
João Alves @joaoqalves.net · 28/10/2025
I've written a sequel to "Disasters I've Seen in a Microservices World," my most-read blog post. Ten disasters later, it's still chaotic, and with the rise of AI agents, it's bound to get worse. Planning to publish soon. Anyone want to read it before I do?
 Four years later, I'm still seeing the same patterns, just dressed in different frameworks, clouds, and YAML dialects. The tools evolved, but the fundamentals didn't: distributed systems remain distributed, humans remain human, and complexity remains undefeated.  What's coming next will make this even more interesting. We're now trying to build AI agents: autonomous, stateful systems that communicate with each other, make probabilistic decisions, and respond to unpredictable inputs. In other words: distributed systems with opinions. What could go wrong? The same fallacies apply, just at a different layer. Latency, consistency, observability, determinism. None of them magically disappears because the component now "thinks."
030
João Alves @joaoqalves.net · 27/10/2025
Sigo contando cosillas :) Hoy inauguro "Desde mi terminal". También en español, pero en formato largo. Explico las razones de crear una cuenta específica para contenido en español. Si queréis leerlo, paso el enlace en el post de abajo. enespanol.joaoqalves.net/p/la-desven...
substack.com
La desventaja de ser plurilingüe
Lo complicado no es hablar varios idiomas, sino decidir en cuál contar tu historia
010
João Alves @joaoqalves.net · 24/10/2025
Everyone talks about multi-cloud and reliability. Here’s what it looked like in practice when I joined Adevinta five years ago. The company had just agreed to acquire the eBay Classifieds Group, its secondhand marketplaces division. On the table: private cloud, GCP, and AWS.
Multi-cloud made it to the final spreadsheet, but only so everyone could see how crazy expensive it would be.  In the past, AWS and other hyperscalers had single points of failure. As @GergelyOrosz  mentions in the latest Pragmatic Engineer newsletter, there are still some components that are “global” and run on a single region (us-east-1). That’s not the norm, and the vast majority of services can be run in multiple areas. It’s up to the business to understand (a) what its critical paths are and (b) what its risk appetite is versus the cost of running multi-region or even multi-cloud/hybrid setups for specific scenarios.  For some businesses, 24/7 uptime is more critical than for others. Take Glovo or Uber, for example: there’s a high per-minute cost during peak hours. Why? The cost of switching is low. Users also have Lyft, FREE NOW, Just Eat, and Uber Eats on their phones, so they can easily open another app to get from A to B or order a meal.  If one of Adevinta’s portals (secondhand, real estate, or motors) is down at 2 a.m. on a Friday, that’s mostly OK. Yes, there are peak times, but there’s also a counterbalance thanks to (a) network effects — a lot of the inventory on @leboncoin  isn’t listed anywhere else; everyone in France uses it — and (b) low time sensitivity. If you’re looking for a flat or a car, it can wait a bit.  Multi-cloud (and sometimes even multi-region) sounds smart until you price it, model the risks, and realize: it’s usually not worth it. Reliability is an architecture problem, not a cloud problem.
000
João Alves @joaoqalves.net · 23/10/2025
Meta laying off a chunk of the FAIR (frontier research AI lab) led by @ylecun was not in my bingo card for 2025. Any clues why? aftermath after acquisition of @alexandr_wang and the team?
000
João Alves @joaoqalves.net · 22/10/2025
At @Adevinta, I didn’t want great work to stay invisible. So I started talking about what we built, encouraged others in 1:1s, and celebrated every post, talk, or shared lesson. “Build and talk” became a habit, and a growth loop. Full article: bit.ly/4nZPWpt
world.hey.com
Karpathy is wrong. Write that post, build that slide deck!
This week, Dwarkesh Patel interviewed Andrej Karpathy and covered the future trajectory of Artificial Intelligence (AI). For those who don't know, Karpathy is a leading AI researcher and co-founder of OpenAI. There's one particular part of the interview that's getting quoted everywhere: People are interpreting that quote as "don't writ...
000