Alessandro Diaferia
@alediaferia.com
Software Engineering Director at Beeline
I publish on themainthread.substack.com.
Featured in “97 Things Every Cloud Engineer Should Know”
https://github.com/alediaferia
I made knotapp.it
I publish on themainthread.substack.com.
Featured in “97 Things Every Cloud Engineer Should Know”
https://github.com/alediaferia
I made knotapp.it
I thought coding agents would make me more productive, but now I just sit back and watch them work
November 5, 2025 at 6:48 AM
I thought coding agents would make me more productive, but now I just sit back and watch them work
Really profound yet concise article, so no point in me articulating much about it: I recommend you read it.
One line that really stuck with me:
“It's easy to create the surface thing, but the right (hard) infrastructure technology compounds.”
amasad.me/keep-winning
One line that really stuck with me:
“It's easy to create the surface thing, but the right (hard) infrastructure technology compounds.”
amasad.me/keep-winning
How to Keep Winning
I've always been fiercely competitive. I enjoy everything about it—from training to get better, to building and developing teams, to the stress, pressure, and intensity. And while I love winning, the....
amasad.me
November 2, 2025 at 8:25 PM
Really profound yet concise article, so no point in me articulating much about it: I recommend you read it.
One line that really stuck with me:
“It's easy to create the surface thing, but the right (hard) infrastructure technology compounds.”
amasad.me/keep-winning
One line that really stuck with me:
“It's easy to create the surface thing, but the right (hard) infrastructure technology compounds.”
amasad.me/keep-winning
Mainly thanks to @antirez.bsky.social’s inspirational videos I’ve gone back to C in my free time and it’s refreshing.
I’m finding joy in developing tiny command line utilities and spending time thinking through details.
I get into the zone instantly and just enjoy the journey.
I’m finding joy in developing tiny command line utilities and spending time thinking through details.
I get into the zone instantly and just enjoy the journey.
April 6, 2025 at 11:54 AM
Mainly thanks to @antirez.bsky.social’s inspirational videos I’ve gone back to C in my free time and it’s refreshing.
I’m finding joy in developing tiny command line utilities and spending time thinking through details.
I get into the zone instantly and just enjoy the journey.
I’m finding joy in developing tiny command line utilities and spending time thinking through details.
I get into the zone instantly and just enjoy the journey.
In this post I collected the traits of what I consider the best software engineers I've been lucky to work with.
Let me know your thoughts, and if I missed some!
open.substack.com/pub/themaint...
Let me know your thoughts, and if I missed some!
open.substack.com/pub/themaint...
January 29, 2025 at 10:33 AM
In this post I collected the traits of what I consider the best software engineers I've been lucky to work with.
Let me know your thoughts, and if I missed some!
open.substack.com/pub/themaint...
Let me know your thoughts, and if I missed some!
open.substack.com/pub/themaint...
With iOS 18 Apple is basically giving the finger to all the people with iPhone 14s or older
January 28, 2025 at 10:15 AM
With iOS 18 Apple is basically giving the finger to all the people with iPhone 14s or older
Am having a lot of fun building an on-call platform from scratch with GenAI’s help.
I picked Go as the main language because I’m familiar with it but haven’t touched it in a few years.
I’m planning to make it all open source soon.
I picked Go as the main language because I’m familiar with it but haven’t touched it in a few years.
I’m planning to make it all open source soon.
January 27, 2025 at 1:47 PM
Am having a lot of fun building an on-call platform from scratch with GenAI’s help.
I picked Go as the main language because I’m familiar with it but haven’t touched it in a few years.
I’m planning to make it all open source soon.
I picked Go as the main language because I’m familiar with it but haven’t touched it in a few years.
I’m planning to make it all open source soon.
I noticed a common flaw of very senior software engineers.
They seem to become increasingly uncomfortable with uncertainty, to the point of stopping making progress until told exactly what to do.
What is your experience? Can you relate?
They seem to become increasingly uncomfortable with uncertainty, to the point of stopping making progress until told exactly what to do.
What is your experience? Can you relate?
January 25, 2025 at 7:51 AM
I noticed a common flaw of very senior software engineers.
They seem to become increasingly uncomfortable with uncertainty, to the point of stopping making progress until told exactly what to do.
What is your experience? Can you relate?
They seem to become increasingly uncomfortable with uncertainty, to the point of stopping making progress until told exactly what to do.
What is your experience? Can you relate?
Some time ago I collected in a post the traits of the most successful software teams I’ve been lucky to work with in my career.
TL;DR it is all about feeling safe to make decisions autonomously and let everyone’s ego aside
themainthread.substack.com/p/best-teams...
TL;DR it is all about feeling safe to make decisions autonomously and let everyone’s ego aside
themainthread.substack.com/p/best-teams...
The best teams I've worked with
The most successful and great to work with teams I've had the pleasure to collaborate with have the common traits that I talk about in this post
themainthread.substack.com
January 22, 2025 at 9:31 AM
Some time ago I collected in a post the traits of the most successful software teams I’ve been lucky to work with in my career.
TL;DR it is all about feeling safe to make decisions autonomously and let everyone’s ego aside
themainthread.substack.com/p/best-teams...
TL;DR it is all about feeling safe to make decisions autonomously and let everyone’s ego aside
themainthread.substack.com/p/best-teams...
I’m having fun designing and exploring concepts I’m less familiar with using LLMs.
I usually have Claude produce markdown docs and flow charts I then include as context in more code-focused conversations to easily create scaffolds (e.g. Copilot)
I usually have Claude produce markdown docs and flow charts I then include as context in more code-focused conversations to easily create scaffolds (e.g. Copilot)
January 20, 2025 at 4:21 PM
I’m having fun designing and exploring concepts I’m less familiar with using LLMs.
I usually have Claude produce markdown docs and flow charts I then include as context in more code-focused conversations to easily create scaffolds (e.g. Copilot)
I usually have Claude produce markdown docs and flow charts I then include as context in more code-focused conversations to easily create scaffolds (e.g. Copilot)
Spending more time planning and designing features is not what helps you avoid regressions and unexpected impact on existing features
It’s cognitive load that’s preventing you from anticipating impact on the existing product
I discuss this aspects and more in my last piece 👇
It’s cognitive load that’s preventing you from anticipating impact on the existing product
I discuss this aspects and more in my last piece 👇
Thoughts on the Liability of the Product Surface
My recent thinking on the impact of ever-growing product surface on the team's ability to deliver value to customers and some recommendations on how to avoid it.
open.substack.com
January 17, 2025 at 2:24 PM
Spending more time planning and designing features is not what helps you avoid regressions and unexpected impact on existing features
It’s cognitive load that’s preventing you from anticipating impact on the existing product
I discuss this aspects and more in my last piece 👇
It’s cognitive load that’s preventing you from anticipating impact on the existing product
I discuss this aspects and more in my last piece 👇
Did you know our working memory can only hold ~7 elements at once? Some say even less.
This limitation is similar to what you might experience when hitting context window limits with GenAI tools. Cont.
This limitation is similar to what you might experience when hitting context window limits with GenAI tools. Cont.
January 16, 2025 at 1:58 PM
Did you know our working memory can only hold ~7 elements at once? Some say even less.
This limitation is similar to what you might experience when hitting context window limits with GenAI tools. Cont.
This limitation is similar to what you might experience when hitting context window limits with GenAI tools. Cont.
A much better capacity allocation approach than the flawed 80-20 split many orgs use is what Abhi proposes in this blog post:
- Strategic work (be it product features or tech initiatives)
- Reactive work (can be product or tech as well)
- Production support work
www.abhishukla.com/elevating-te...
- Strategic work (be it product features or tech initiatives)
- Reactive work (can be product or tech as well)
- Production support work
www.abhishukla.com/elevating-te...
Elevating Technical Work Into Investment: Escaping the 20% Trap
www.abhishukla.com
January 15, 2025 at 9:54 AM
A much better capacity allocation approach than the flawed 80-20 split many orgs use is what Abhi proposes in this blog post:
- Strategic work (be it product features or tech initiatives)
- Reactive work (can be product or tech as well)
- Production support work
www.abhishukla.com/elevating-te...
- Strategic work (be it product features or tech initiatives)
- Reactive work (can be product or tech as well)
- Production support work
www.abhishukla.com/elevating-te...
I have been collecting some thoughts on the implications of ever-growing software product surface on development teams.
There's plenty of related research out there on the topic, which I tried to link throughout my article.
If it resonates with you, check out alediaferia.com/2025/01/14/p...
There's plenty of related research out there on the topic, which I tried to link throughout my article.
If it resonates with you, check out alediaferia.com/2025/01/14/p...
Thoughts on the Liability of the Product Surface
Discover how unmanaged product surface growth silently undermines software teams' effectiveness. Learn why treating every feature as an asset might be limiting your ability to deliver value.
alediaferia.com
January 14, 2025 at 12:04 PM
I have been collecting some thoughts on the implications of ever-growing software product surface on development teams.
There's plenty of related research out there on the topic, which I tried to link throughout my article.
If it resonates with you, check out alediaferia.com/2025/01/14/p...
There's plenty of related research out there on the topic, which I tried to link throughout my article.
If it resonates with you, check out alediaferia.com/2025/01/14/p...
I like to think of the software we deliver as moving on an asset/liability spectrum over time.
As the conditions change it’s our responsibility to get rid of high-liability pieces of that software.
It also helps keeping cognitive load down
As the conditions change it’s our responsibility to get rid of high-liability pieces of that software.
It also helps keeping cognitive load down
January 3, 2025 at 12:35 PM
I like to think of the software we deliver as moving on an asset/liability spectrum over time.
As the conditions change it’s our responsibility to get rid of high-liability pieces of that software.
It also helps keeping cognitive load down
As the conditions change it’s our responsibility to get rid of high-liability pieces of that software.
It also helps keeping cognitive load down
There is only a finite amount of context we can keep in our head when doing creative knowledge work.
It’s like a limited mental scratchpad.
We shouldn’t estimate this when sizing teams around domains: how much knowledge can they sustain to remain effective?
It’s like a limited mental scratchpad.
We shouldn’t estimate this when sizing teams around domains: how much knowledge can they sustain to remain effective?
January 3, 2025 at 11:27 AM
There is only a finite amount of context we can keep in our head when doing creative knowledge work.
It’s like a limited mental scratchpad.
We shouldn’t estimate this when sizing teams around domains: how much knowledge can they sustain to remain effective?
It’s like a limited mental scratchpad.
We shouldn’t estimate this when sizing teams around domains: how much knowledge can they sustain to remain effective?
Came across this interesting article on measuring cognitive load.
I argue engineering organizations do not prioritize enough reducing cognitive load, and rather default to doing the opposite (generally implementing more features and hardly cutting them back)
docs.kedehub.io/kede/kede-on...
I argue engineering organizations do not prioritize enough reducing cognitive load, and rather default to doing the opposite (generally implementing more features and hardly cutting them back)
docs.kedehub.io/kede/kede-on...
How to measure developer happiness?
A knowledge-centric approach
docs.kedehub.io
January 2, 2025 at 8:35 AM
Came across this interesting article on measuring cognitive load.
I argue engineering organizations do not prioritize enough reducing cognitive load, and rather default to doing the opposite (generally implementing more features and hardly cutting them back)
docs.kedehub.io/kede/kede-on...
I argue engineering organizations do not prioritize enough reducing cognitive load, and rather default to doing the opposite (generally implementing more features and hardly cutting them back)
docs.kedehub.io/kede/kede-on...
I’m experimenting with adding illustrations to my blog post.
With this one I’m trying to convey some of the many type of features that contribute to the whole “product surface”
With this one I’m trying to convey some of the many type of features that contribute to the whole “product surface”
December 30, 2024 at 3:17 PM
I’m experimenting with adding illustrations to my blog post.
With this one I’m trying to convey some of the many type of features that contribute to the whole “product surface”
With this one I’m trying to convey some of the many type of features that contribute to the whole “product surface”
If your team is struggling to justify work on tech debt, or if you have a fixed allocated capacity for tech debt work (I bet it's 20%) then your organization is probably paying attention to the wrong metrics.
You will hardly have the chance to remove unused code because throughput is key. 1/3
You will hardly have the chance to remove unused code because throughput is key. 1/3
December 30, 2024 at 11:54 AM
If your team is struggling to justify work on tech debt, or if you have a fixed allocated capacity for tech debt work (I bet it's 20%) then your organization is probably paying attention to the wrong metrics.
You will hardly have the chance to remove unused code because throughput is key. 1/3
You will hardly have the chance to remove unused code because throughput is key. 1/3
Cognitive load significantly impacts your team’s ability to deliver.
Too much stuff to keep in mind slows you down.
Imagine designing a new feature but having to keep in mind a bunch of corner cases that you must not break.
Exhausting
Too much stuff to keep in mind slows you down.
Imagine designing a new feature but having to keep in mind a bunch of corner cases that you must not break.
Exhausting
December 29, 2024 at 3:22 PM
Cognitive load significantly impacts your team’s ability to deliver.
Too much stuff to keep in mind slows you down.
Imagine designing a new feature but having to keep in mind a bunch of corner cases that you must not break.
Exhausting
Too much stuff to keep in mind slows you down.
Imagine designing a new feature but having to keep in mind a bunch of corner cases that you must not break.
Exhausting
If you are a language nerd like me you are going to love this www.ted.com/dubbing/lera...
How language shapes the way we think
There are about 7,000 languages spoken around the world -- and they all have different sounds, vocabularies and structures. But do they shape the way we think? Cognitive scientist Lera Boroditsky shar...
www.ted.com
December 29, 2024 at 3:11 PM
If you are a language nerd like me you are going to love this www.ted.com/dubbing/lera...
I'm building this framework that provides a DSL for defining entities, actors, operations and a VM that enforces those definitions.
Main goal is to help managing complexity of software project.
I know, it's naïve to even try. But am having fun.
Main goal is to help managing complexity of software project.
I know, it's naïve to even try. But am having fun.
December 28, 2024 at 8:46 AM
I'm building this framework that provides a DSL for defining entities, actors, operations and a VM that enforces those definitions.
Main goal is to help managing complexity of software project.
I know, it's naïve to even try. But am having fun.
Main goal is to help managing complexity of software project.
I know, it's naïve to even try. But am having fun.
This is something product teams should read once every quarter to say the least hackernoon.com/12-signs-you... by @johncutle.fish
HackerNoon - read, write and learn about any technology
How hackers start their afternoon. HackerNoon is a free platform with 25k+ contributing writers. 100M+ humans have visited HackerNoon to learn about technology
hackernoon.com
December 27, 2024 at 3:01 PM
This is something product teams should read once every quarter to say the least hackernoon.com/12-signs-you... by @johncutle.fish
When product teams lack discipline on validating that what they build is delivering value, they will hardly have the chance to clean up tech debt after themselves, ultimately becoming slower and more error prone over time.
1/3
1/3
December 27, 2024 at 2:44 PM
When product teams lack discipline on validating that what they build is delivering value, they will hardly have the chance to clean up tech debt after themselves, ultimately becoming slower and more error prone over time.
1/3
1/3
Main ways I'm taking advantage of LLMs at the moment:
- Rubberducking: getting immediate feedback on ideas, mainly exploring aspects I'm less familiar with
- Writing additional test cases to what I already have
- Refactoring, simplifying code
- Rubberducking: getting immediate feedback on ideas, mainly exploring aspects I'm less familiar with
- Writing additional test cases to what I already have
- Refactoring, simplifying code
December 23, 2024 at 1:33 PM
Main ways I'm taking advantage of LLMs at the moment:
- Rubberducking: getting immediate feedback on ideas, mainly exploring aspects I'm less familiar with
- Writing additional test cases to what I already have
- Refactoring, simplifying code
- Rubberducking: getting immediate feedback on ideas, mainly exploring aspects I'm less familiar with
- Writing additional test cases to what I already have
- Refactoring, simplifying code