Your Team's AI Spend Is a Black Box - Even to You
Every engineering leader can quote the token bill. Here's how to tie it to what actually got built - and use that in your next call with product.
Filipe Albero Pomar · London Hi, I'm Filipe.
I'm an engineering manager who's also been the product manager and the architect.
Today I'm at Maersk. Before that, Sky News and Now TV. Along the way I've become hands-on with AI, and I write and speak about what it means for engineering teams.
10 yrsin engineering management, after years as an engineer
3→15engineers, twice. Teams I built from the ground up
C-levelbriefed CPOs and CTOs at three companies
+4%annual recurring revenue from one product I designed and built
Where I've worked
- Maersk
- Sky News
- Now TV
- plus detours into energy trading and virtual reality
01
A bit more about me
The short version, if you'd asked me to tell you about myself.
I care about two things in equal measure: software that moves the business, and the people who build it. Software with no business impact means nothing, and a team that doesn't work well together won't deliver it.
When Apple News launched, its technical constraints threatened to leave an entire newsroom on the sidelines. I found a way in. One product I designed and built added 4% to annual recurring revenue.
The less visible part of the job is individuals, one at a time, and then the dynamics between them. That's the part I enjoy most.
How I got to London
- Brazil grew up
- Alaska school exchange year
- Brazil BSc Computer Science
- Italy MSc Software Engineering
- Spain MSc, continued
- London home since 2011
02
What I do
Engineering manager, sometime product manager, still close to the code. In practice, that means four things.
People
Growing engineers
Career coaching and progression, one person at a time, then the dynamics between them.
Technology
Setting technical direction
I act as architect for my teams, shaping the systems and the trade-offs, including how we build with AI.
Product
Building the right thing
Working with product to sharpen what we build before we build it. I've done the PM job too.
Delivery
Shipping it
Organising delivery across the UK , the Netherlands and India . I've also briefed CPOs and CTOs on progress and risk.
> what I believe
AI is on its way to being a commodity. When everyone has the same tools, the real advantage is the people, and how well they work together.
It's the thread running through everything I speak and write about.
03
Talks and writing
I write and speak about how AI is changing engineering teams, and the parts of management that rarely show up in the metrics. Some recent pieces:
Articles
What Engineers Miss About Promotion Decisions
Most engineers think better code means faster promotion. Their manager is watching something else entirely.
Your AI Productivity Number Is Right. So Is Your Team's
What to do when your data and your team tell you different things.
Talks
-
GenAI London Institutional Memory: Turning AI Coding Assistants into a Team's Collective Brain
-
TechLeadConf, London Cross-Team AI Application
-
LinkedIn Live, TechLeadConf warm-up The Vanishing Manager - How Many Managers Does a Team Even Need?
04
Kind words
I'm grateful to have received these comments from past colleagues.
Some say that your first manager has a major impact on your career progression. Well, if my career goes just half as well as Filipe was as my first proper manager, I have a bright future ahead of me.
Filipe is a consummate professional who had the respect of both his team and senior stakeholders alike.
Filipe is very adaptable and open to learning new things. He is approachable, patient and focused on the needs of the teams he works with. He has a positive attitude and looks for ways he can drive impact.
05
Still curious? Ask my AI Twin.
That 90s-style terminal docked at the bottom of the screen is my AI Twin. It answers questions about my career the way I would, drawing on my CV and career history.
I built it to showcase the same AI skills my teams use every day. It runs on Cloudflare Workers and Gemini. Automated evals, with a second AI as the judge, check its answers for accuracy, tone and security.