Performance at scale with Ryan Townsend
New episode of Señors @ Scale is out. Ryan Townsend is Principal Product Manager for Performance at Cloudflare, after 20+ years building for the web and a decade as a CTO.
We get into why "Rails doesn't scale" was always a myth, how he cut CI from 40 minutes to 3, the "no performance impact" personalization script that cost seconds at P90, and what it takes to hit a 99%+ cache hit ratio before Black Friday.
Watch the full episode on YouTube
Listen on Spotify
Community reads
Performance
How we made claude.ai 3x faster in two weeks by Raymond Wang, Sam Attard, and Issac G. — users kept saying it was slow and they were right, so the team ran the whole sprint from one Slack channel with Claude in every thread.
Optimizing a Lock-Free Ring Buffer by David Álvarez Rosa — from 12M ops/s with a mutex to 305M, one step at a time.
The Cloudflare Blog — brought to you by EmDash by Kody Jackson, Diogo Carneiro, and Amy Dutton — they moved their own blog onto their own CMS to prove it holds up, and a 28,000 RPS DDoS attack showed up to help with the test.
AI and agents
The Code Nobody Reads by Addy Osmani — two years ago Addy's own tip was to review every line of generated code, so what earns your trust now that nobody reads it all?
Coding is NOT solved by Alex Ewerlöf — writing code got cheap, and anyone who has run software in production knows which part of the bill that was.
Introducing Clef: our open-source decision models, and new RL fine-tuning platform by Michelle Chen, Alex Reneau, and Kevin Flansburg — decision models are the buzz of the last few weeks, so what are they and when does an agent want one over an LLM?
The Bitterest Lesson by Diogo — Sutton's bitter lesson says compute beats clever algorithms, yet a GPT-2-sized model over 100x smaller than GPT-3 destroyed it.
Tooling
Effect 4.0 by Sebastian Lorenz — the minimal bundle dropped from 35.6 kB to 7.1 kB.
We want you to build the next Git platform on Cloudflare by Dina Kozlov, Dillon Mulroy, and Zebulon Piasecki — Cloudflare is running a competition to build the GitHub for agents (submissions close October 14).
Design
how design gets done in 2026, part 3 / inside the AI infrastructure of Stripe, Fin (formerly Intercom), Anthropic, Vercel + Linear by Ioana from AI Goodies — five design orgs, including the Linear designer who caps AI at 20 to 30% of new work on purpose.
Career
I ___ Netlify by Eduardo Bouças — the dishwasher wiped the heart off an "I ♥ Netlify" mug, and now the gap needs a word.
Barcelona AI Hackathon
I'm hosting a hackathon in Barcelona on October 9 and 10.
Teams of up to four get the theme on Friday evening and demo on Saturday afternoon. The best build wins €10,000 in cash.
There are around 25 spots left out of 70, so if you're in Barcelona, apply.
Apply for the Barcelona AI Hackathon
Book recommendation

Scale: The Universal Laws of Growth, Innovation, Sustainability, and the Pace of Life in Organisms, Cities, Economies, and Companies by Geoffrey West
West is a physicist and the former head of the Santa Fe Institute, and he found that a mammal's size predicts its heart rate and how long it lives. In the book he takes the same scaling laws to cities and then to business, to work out why some companies last and others don't.
Find it on Goodreads
Meme of the week

Conferences
CityJS Athens
CityJS lands in Athens this October.
My link takes 10% off the ticket price for you.
Buy tickets
That's it for this one.
If you're enjoying these, share the subscribe link with someone who'd get something out of it, or with your team on Slack:
neciudan.dev/subscribe