Pelayo Arbués

@pelayoarbues.com

6681 Followers 1197 Following

Head of Data Science and AI at idealista.com | Film photographer | Ph.D. in Economics | Sharing my digital garden at pelayoarbues.com | Working full remote from the North of Spain

idealista map showing the new feature
Photography of the book Tres Millones de viviendas displayed on top of orange bricks
A text editor in baroque style, created using magnific.ai
A digital landscape showing gradient descent
Evolution from character to person. When we don’t know someone well, we tend to project intentions onto them, casting them as a character in a play they themselves are unaware exists. It’s quite challenging to optimize on behalf of characters in your mental play, but it’s much easier to be understanding of people you know personally. Spending time together learning about each other, often at a team offsite, will slowly transform strangers into people.
AI generated image of a bridge linking two lands
In simple terms, the purpose of data is to give you a causal model of your business in your head. If reality changes, the figures in your business should show it. At which point you get to update your knowledge.
Coaching is no longer a specialty; you cannot be a good manager without being a good coach. The path to success in a fast- moving, highly competitive, technology- driven business world is to form high- performing teams and give them the resources and freedom to do great things.
When leaders take steps to protect their own interests, particularly when those steps are taken at the expense of others, they send a message to everyone else that it is okay to do the same.
To maintain the core of your product there are usually one or two things that have to stay still while everything else spins and changes around them. And that’s a useful constraint. You need some constraints to force you to dig deep and get creative, to push envelopes you hadn’t thought to open before
Yes, leadership is about vision. But leadership is equally about creating a climate where the truth is heard and the brutal facts confronted
If people only comply, we can’t expect people to take responsibility for their actions. The chain of command is for orders, not information. Responsibility is not doing as we are told, that’s obedience. Responsibility is doing what is right.
The push for commercial control of AI capability is dangerous. Naomi Klein, who coined the term “shock doctrine” as “the brutal tactic of using the public’s disorientation following a collective shock… to push through radical pro-corporate measures”, is now warning that AI is “likely to become a fearsome tool of further dispossession and despoilation”.
Having two different pipelines to process your data is a common cause for bugs in ML production. One cause for bugs is when the changes in one pipeline aren’t correctly replicated in the other, leading to two pipelines extracting two different sets of features. This is especially common if the two pipelines are maintained by two different teams, such as the ML team maintains the batch pipeline for training while the deployment team maintains the stream pipeline for inference
Every three to six months, audit your meetings. Are they useful? Are the right people involved? Do they achieve their purpose, and are they worth the time? Could more be accomplished asynchronously? Consider polling participants or having a neutral person do so.
Not finance. Not strategy. Not technology. It is teamwork that remains the ultimate competitive advantage, both because it is so powerful and so rare.
Make sure your candidates know that you’re excited about them. I first encountered this idea reading Rands’s “Wanted” article, 4 and he does an excellent job of covering it there. The remarkable thing is how few companies and teams do this intentionally: in my last interviewing process, three of the companies I spoke with expressed interest exceptionally well, and those three companies ended up being the ones I engaged with seriously.
Self-awareness has three components: understanding your underlying value system, identifying your innate preferences—your work style and decision-making tendencies—and being clear about your own skills and capability gaps.
To write something, you have to do things, collect ideas, process them, put words on the page, structure them, edit them, rewrite them, and publish them to your site — and that’s all for one article. Then, there’s the emotional challenges: writer’s block, imposter syndrome, lack of time, judgment from strangers, and criticism from friends.
Any given executive is almost always uncannily good at one way of consuming information. They feel most comfortable consuming data in that particular way, and the communication systems surrounding them are optimized to communicate with them in that one way.
I would argue operational clarity is exactly the opportunity many data teams have been looking for. So many leaders we speak to say something like “We can do so much more.” Adding operational clarity is something undeniably valuable that data folks are uniquely positioned to do well.
and the pressure to compete on things like price, and other short- term strategies, goes up.
I once read somewhere on the internet: deploying is easy if you ignore all the hard parts
Being able to set ego aside and be ruthlessly critical of yourself and your division is hard. It’s not just about where you have gaps today but also about what you’ll need in the next three to five years.
Screenshot of the quote and cover of book
Screenshot of prompt: Act like a very experienced Chief Data Scientist. Design a system for tackling fraud in an online marketplace similar to Zoopla.
An illustration made with AI depicting a beast in a futuristic library
We think we can easily see into the hearts of others based on the flimsiest of clues. We jump at the chance to judge strangers. We would never do that to ourselves, of course. We are nuanced and complex and enigmatic. But the stranger is easy.
Robots pictured in a decision tree
Stakeholders will sometimes pressure you to find data that supports a narrative they have already created in advance. While playing along with this might score you some points in the near term, what will help you in the long term is being a truth seeker and promoting the narrative that the data truly supports
Instead of condemning people, let’s try to understand them. Let’s try to figure out why they do what they do. That’s a lot more profitable and intriguing than criticism; and it breeds sympathy, tolerance and kindness. “To know all is to forgive all.”
Companies that bubble- wrap their employees with tons of free perks are usually shortsighted and have no long- term strategy to sustain those perks, or they have an innately problematic core business and the perks are the cover.
Give a good idea to a mediocre team, and they will screw it up. Give a mediocre idea to a great team, and they will either fix it or come up with something better. If you get the team right, chances are that they’ll get the ideas right.
We stopped talking about building teams, and talked instead of growing them. The agricultural image seemed right. Agriculture isn’t entirely controllable. You enrich the soil, you plant seeds, you water according to the latest theory, and you hold your breath. You just might get a crop; you might not.
You must maintain unwavering faith that you can and will prevail in the end, regardless of the difficulties, AND at the same time have the discipline to confront the most brutal facts of your current reality, whatever they might be.
“Don’t assume the folks who got you here will get you there.” Be constantly vigilant in answering this question: Do I have the talent I need now and for the next two to four years for my team or company to be a success?
It’s poison to think great ideas can only come from you. That you alone can hoard them in one place. And it’s stupid. Wasteful. A CEO has to recognize fantastic ideas— regardless of their source.
management technique he called the “rule of two.” He would get the two people most closely involved in the decision to gather more information and work together on the best solution, and usually they would come back a week or two later having decided together on the best course of action.
The book A manual of design fiction and a notebook on it
A stranded bottle on a Galician beach.
Quote: "The propensity to lead without being given the authority to do so is what, in organizations, distinguishes people that can innovate and break free of the constraints that limit their competitors. Innovation is all about leadership, and leadership is all about innovation. The rarity of the one is a direct result of the rarity of the other."
Quote from book: "Because ML systems are part code, part data, and data can change quickly, ML systems need to be able to evolve quickly. This is tightly linked to maintainability."
Nerea Luis speaking about AI on a stage
A woman pouring Asturian apple cider in the traditional style.
Quite from the book Build: What you’re building never matters as much as who you’re building it with.
A clear blue sky over Barcelona highlights a 60-meter-tall column topped with a bronze statue of Christopher Columbus.
Quote from Leaders eat last book: Cooperation does not mean agreement, i means working together to advance the greater good
View of the Prepyrenees mountains between France and Andorra
Covent Garden Christmas decorations
AI generated illustration of a desktop with a laptop
Screenshot of idealista app on dark mode
Quote from the book Leaders eat last: Koppel says that news organizations went from delivering the news you need, even if you don’t want it, to the news you want even if you don’t need it."
AI generated image of a minimalist room with a great library
Quote from the book Trillion dollar coach: "If you are a manager, executive, or any other kind of leader of teams, in any kind of business or organization, you can be more effective and help your team perform better (and be happier) by becoming the coach of that team."
Power balance wristband