Derek (@derek_arg) header
Derek (@derek_arg) profile picture

edge infra @airbnb, @envoyproxy contributor, prev. @tesla and @pinterest 🇩🇴🇸🇻

Austin, TX
Joined April 2014
Photo by Derek (@derek_arg)Photo by Derek (@derek_arg)Photo by Derek (@derek_arg)Photo by Derek (@derek_arg)Photo by Derek (@derek_arg)Photo by Derek (@derek_arg)Photo by Derek (@derek_arg)Photo by Derek (@derek_arg)Photo by Derek (@derek_arg)
A lovely stack overflow answer to "why use C++ over C", particularly the ending. I agree that there's no language that shields you from misuse. Spaghetti Python is still a nightmare to refactor and I've seen (and written) some pretty ugly Go softwareengineering.stackexc…
1
13
Derek retweeted
Software Engineering at the Tipping Point A deep look at the systemic effects of a tenfold shock on developer ecosystems Software systems remain inherently human-centric. Human judgment, core engineering values, and strategic adaptation dictate success queue.acm.org/doi/10.1145/38…
1
6
441
* no more 3rd-party dependencies for small stuff, not worth supply chain risk to use a 50-line library from an internet rando * principle of least privilege everywhere * mTLS everywhere * continuously fuzz-test all interfaces * sign all artifacts * 2FA * DDoS protection * pray
1
2
80
dammit ya’ll
37
let's goooooo
1
4
167
off by 1 minute :/
17
automated testing infrastructure still matters
rust DOES NOT guarantee bug free code!! This new wave of my agent is writing rust so it's obviously the perfect correct code is making me so mad! Please just fucking understand things for yourself
48
If coding agents are making mistakes then there's not enough "engineering" in the software engineering
1
1
48
user error: it was compression
i have a deployment of nginx’s that locally forward to envoy (i know). despite nginx doing TLS termination & client connection handling, nginx’s CPU usage is like 1/2 of envoy. sooo i’m poking around to see what i can squeeze out of envoy github.com/envoyproxy/envoy/…
1
147
i have a deployment of nginx’s that locally forward to envoy (i know). despite nginx doing TLS termination & client connection handling, nginx’s CPU usage is like 1/2 of envoy. sooo i’m poking around to see what i can squeeze out of envoy github.com/envoyproxy/envoy/…
2
1
2
297
i know nginx is highly optimized C but they say C++ has a “zero-overhead principle” so let’s C about that
1
39
TEXAS IS SO FUCKING BACK, Y'ALL
32
113
2
1,588
14,035
good thing no one is still using ingress-nginx. right?
73
Taylor Swift's silence on the PHP CVE is deafening
106
1,659
102
20,484
345,766
Replying to @vineerpasam
AI made writing code cheap, but good engineering matters even more now.
1
1
72
6,983
Give your agent a way to verify its work. This can dramatically improve the quality of output. With this process in place, Claude Code was able to write 90%+ of its code itself. When generating software at a faster rate than you can review it, the bottleneck shifts to proving that it works. @bcherny and team invested enough in verification, we've been able to move fast while maintaining stability. Code reviews aren't going away, but you need to think of it differently. One human reading through line-by-line is no longer going to work as a gate, it needs to evolve into a multi-layered verification system, built with fast deterministic feedback loops for your agents. This is where e2e tests, feature maps (as @poteto has been advocating), property tests and multi-agent adversarial reviews come in. These all provide opportunities to catch errors early and fix them before things get out of hand. You don't have enough eyeballs for the amount of code being written so your focus needs to be on product intent, risk assessment and evaluating evidence rather than reviewing code. Delegate the code checking work to your automated tests, scale your approvals based on risk, and turn your test failures into living evaluations. This approach helps maintain speed without losing reliability. Let the agents do what they do best and let you focus on what makes you irreplaceable: judgement, direction, and ownership. We tried to capture how we handle code review and verification in this writeup. Hopefully it's helpful.
Claude writes about 90% of Claude Code's code. From June to September, merged PRs more than doubled while the share reverted fell by more than 40%. We wrote about how we changed code review to get there, and how to adopt it: claude.dev/blog/solving-code…
220
176
61
2,534
200,948
tbh I don’t see the irony here, I think 2 engineers using the same model does not mean the same performance/output. My Opus’ skills and context make it a better engineer than anyone else’s Opus that i’ve seen.
90% of software engineering in 2026: Engineer 1 → AI: “Build this and open a PR” Engineer 1 → Engineer 2: “Can you review?” Engineer 2 → AI: “Review this PR” Engineer 2 → Engineer 1: “LGTM” The irony: both engineers use the exact same AI model.
1
64
i had my claude read Effective Go. In effect, it writes pretty effective Go.
To create my claude.md, I had it read: - my favorite programming books + blogs - the best oss projects - the code of the engineers I respect the most Claude (when given proper instructions, constraints, and context) now writes better code than me, almost always.
1
65
Derek retweeted
Rust compiler: "I cannot allow you to compile this because reference x might outlive borrowed scope y on line 42." C compiler: "You want to cast an integer to a function pointer and jump straight into uninitialized heap memory? Hell yeah brother, here's your binary."
105
448
54
10,168
276,796