Most software teams don’t ship badly because they lack talent.

In this week’s Signal vs Noise, I break down a boring-but-effective shipping playbook that makes releases feel routine instead of heroic.

Subscribe for practical takes on engineering that actually scale!

substack.com/home/post/p-...

AI is an implementation detail not a value prop.

At the end of the day, customers buy outcomes not the models that drive them.

If you can't sell the pen without saying “AI,” you probably don't understand the problem well enough yet.

Defining what you're actually trying to get done is step #1.

💻 Software Engineering 101: Learning Through Memes

Today’s Lesson: Lambda

The promise of AWS Lambda is simple: Upload a function. An event triggers it. You pay per execution

But cold starts, harder debugging, and unexpected costs can all cause issues

What’s your best (or worst) Lambda story?

AI was 2025’s favorite promise. 2026 is the year engineering leaders have to live with the consequences. Most teams now feel more productive with AI but the data is blunt. Developers are faster on tasks while code churn, security incidents, and “almost‑right” fixes are climbing.

AI didn’t “break” software development in 2025, it exposed where our incentives were already weak. AI is a powerful tool. But in the wrong incentive system, it just helps us make a mess faster.

This week on Signal vs Noise, I write about why pass/fail thinking collapses the moment you ship probabilistic systems and how eval fills the gap that testing was never designed to handle.

signalvsnoisetechinsights.substack.com/p/ai-evals-a...

AI won’t replace developers but it will divide teams that learn to manage it from those that blindly adopt it. Leadership’s job in 2026 is to design workflows, guardrails, and skills so people can use these tools with judgment.

💻 Software Engineering 101: Learning Through Memes

Today’s Lesson: Moore’s Law

Moore’s Law isn’t physics. It's an observation made by Gordon Moore in 1965: transistors roughly doubled every ~2 years

Now Moore's law is slowing as transistor are harder to shrink as we run into physical limitations.

If your interview process doesn't cover salary, responsibilities, and work culture, what exactly do you talk about?

There are two kinds of companies.

Type 1: It’s good when lots of people know who you are
👉 Visibility = opportunity
Think startups

Type 2: It’s bad when lots of people know who you are
👉 Visibility = risk
Think large corporations

Figure out which environment you’re in and optimize accordingly.

MongoDB didn’t fail... but at some point, it did stop being the default.

This week on Signal vs Noise, I break down what actually happened and why it matters beyond MongoDB.

👉 Signal vs Noise: weekly, no-BS engineering strategy

signalvsnoisetechinsights.substack.com/p/whatever-h...