- Pointer
- Posts
- Issue #684
Issue #684
Essential Reading For Engineering Leaders
Friday 23rd January issue is presented by Upsun
Scaling isn’t about traffic spikes. It’s about removing fragile steps before they break production. In this case study, see how Yard handled:
25M+ requests in production
Zero-downtime deployments
Predictable scaling without infra rewrites
A real-world look at reliability under pressure.
— Lara Hogan
tl;dr: “So what do you do when you’re working with a teammate who is stuck in a cycle of unhelpful or unproductive behavior? You’ve got empathy for them; you don’t want to be a jerk about it. But you still need this person to change course and start moving forward.”
Leadership Management
— Wes Kao
tl;dr: “Vague language can have a large (negative) blast radius when used haphazardly. It can create anything from avoidable anxiety to hours spent going in the wrong direction due to miscommunication. Yet, it’s surprisingly common in the workplace. Most of us have used vague language at work, and have been on the receiving end of it.”
Leadership Management
tl;dr: Scaling isn’t about traffic spikes. It’s about removing fragile steps before they break production. In this case study, see how Yard handled: (1) 25M+ requests in production. (2) Zero-downtime deployments. (3) Predictable scaling without infra rewrites. A real-world look at reliability under pressure.
Promoted by Upsun
Management Cloud
— Joel Hawksley
tl;dr: “The senior role is a crossroads. At many companies, it’s the farthest you can go as an individual contributor. You basically have one option for promotion: management. But it doesn’t have to be this way. Your journey as an individual contributor doesn’t have to stop at senior. There is sometimes another option: staff.”
Leadership Management CareerAdvice
"Leadership is the art of giving people a platform for spreading ideas that work."
— Gergely Orosz
tl;dr: “Crafting well-structured diagnostics with useful messages is an incredibly valuable and high-leverage way to spend your time. For many applications and platforms with complex and open-ended inputs, diagnostics are the primary interface - the vast majority of the user’s time will be spent dealing with errors and progressing to the next one.”
Guide
— Kyle Galbraith
tl;dr: Kyle Galbraith, CEO of Depot, shares insights from processing 100 million builds in 2025 and saving teams 968 years of build time. Learn why external systems can be the biggest scaling challenge, how data becomes infrastructure, and why AI coding agents mean the bottleneck has shifted from writing code to building and shipping it faster.
Promoted by Depot
Platform Tools
— Ethan McCue
tl;dr: “There is a set of things that you can do when working with a Postgres database which I have found made my and my coworker's lives much more pleasant. Each one is by itself small, but in aggregate have a noticeable effect.”
PostgreSQL
— Steve Klabnik
tl;dr: “I’m going to warn you about this post: I hope that by the end of it, you’re a little frustrated. This is because I don’t think it’s productive to skip to the tools and techniques that experienced users use yet. We have to walk before we run. And more importantly, we have to understand how and why we run.”
Agents AI
— Matthias Endler
tl;dr: “I’ve been reviewing other people’s code for a while now, more than two decades to be precise. Nowadays, I spend around 50-70% of my time reviewing code in some form or another. It’s what I get paid to do, alongside systems design. Over time, I learned a thing or two about how to review code effectively. I focus on different things now than when I started.”
CodeReview Guide
Null Pointer

Cost-Optimized
Hand-drawn by Manu. Got an idea for a cartoon? Click reply and let us know
Most Popular From Last Issue
Three Ways To Solve Problems - Andreas Fragner
Notable Links
ChartGPU: OS WebGPU-based charting library.
Clawdbot: Personal AI assistant on your own devices.
Eigent: OS cowork desktop to unlock productivity.
Nanolang: Experimental language to be targeted by coding LLMs.
Swark: Create architecture diagrams from code automatically using LLMs.
How did you like this issue of Pointer?1 = Didn't enjoy it all // 5 = Really enjoyed it |
