Casey Chow

142 posts

Casey Chow

Casey Chow

@chasey_cow

New York Katılım Ağustos 2010
102 Takip Edilen86 Takipçiler
Casey Chow retweetledi
Yijie
Yijie@yijiefeng·
I'm noticing a trend there's a growing number of "AI consulting" firms charging $20K+ to "deploy Claude" to legacy businesses (mid-sized law, accounting, PE firms) as "Anthropic enterprise partners" what this means: - they install claude code, cowork - run a few commands to connect to tools - give generic or misleading advice on a tech stack meanwhile, there's more interest than ever for firms with 0 technical staff to build SaaS in-house and there's an entire industry of advice givers seeking to profit off of this trend last week I was on a call with a PE firm (working with one of these agencies) and someone who had never written code was asking whether to run a RAG vector DB on a Mac mini to chunk internal docs can't make this up
English
129
51
1.6K
249.2K
Casey Chow
Casey Chow@chasey_cow·
@JoshPurtell @_lopopolo What’s “better” here? I wouldn’t be surprised for any one benchmark, but over many metrics that we still care about is far more interesting.
English
1
0
0
70
Josh
Josh@JoshPurtell·
@_lopopolo Can I operationalize this? Would you be surprised if I optimized a scaffold with gpt-5-mini such that it does better in it than in the codex version that was out then, and then compared both harnesses with gpt-5.4-mini and the non-codex harness did better?
English
3
1
13
1.2K
Ryan Lopopolo
Ryan Lopopolo@_lopopolo·
While alternative coding harnesses may have short term lift, they will be bitter lesson’d away. I am bearish on any harness that doesn’t come from the lab whose model you are using. You’re fighting against post-training. To put a finer point on this, you know how like, ioctls are like “huh that's weird but I guess whatever it's what we've got we can work with that”? It is exact the same with like, the particular JSON construction the Codex shell tool uses. The model used to mangle nested quotes in this monstrosity RPC all the time but now it does not and it does not matter that the API is bad because billions of failed invocations are used to train to the harness we have, not the harness we deserve.
English
76
19
297
107K
Casey Chow retweetledi
Linear
Linear@linear·
Introducing Linear Releases. Manage software releases directly from Linear. Track the deployment environment, version, and status of every issue to give team members and agents your full deployment context.
English
50
86
1.8K
253K
Casey Chow retweetledi
Yunyu Lin
Yunyu Lin@yunyu_l·
Working only 1 job is low signal in 2025. Do you have what it takes to be the next Soham Parekh? Introducing Startup Interview Simulator: convince (simulated) founders to hire you, and secure the biggest bag possible.
English
31
25
259
75.8K
Casey Chow
Casey Chow@chasey_cow·
On traditional Rails, that makes total sense. A rails instance could only serve one request at a time. But node can handle hundreds or thousands of concurrent requests. So 28 instances means the instance event loops are just crashing into each other instead.
English
0
0
0
256
Casey Chow
Casey Chow@chasey_cow·
Biggest example for me: turns out Node.js is more CPU bound than memory bound. Performance L instances give you 14GB of RAM but only 4 vCPUs. If you follow the default advice, you cluster 28 Node instances to handle requests.
English
1
0
1
109
Casey Chow
Casey Chow@chasey_cow·
So after going through and practicing most of Heroku’s documentation over the past few years, it turns out most of their their scaling advice is just plain wrong if you’re not running Ruby on Rails (or maybe Python, but I know less about that).
English
1
0
0
208
Casey Chow
Casey Chow@chasey_cow·
@mscccc Alternatives considered: refactoring the entire backend in APL
English
0
0
1
75
Mike Coutermarsh
Mike Coutermarsh@mscccc·
Please show me the Architecture Decision Record™️ where it was decided we’d write 200k lines instead of 20k lines
English
1
0
19
3.4K
Casey Chow retweetledi
Rob Leathern
Rob Leathern@robleathern·
Oh yes you can. “Create a drinking game based on common product management mistakes”
Rob Leathern tweet media
English
7
27
123
0
Casey Chow retweetledi
Gergely Orosz
Gergely Orosz@GergelyOrosz·
I'm observing more and more startups choose a 'TypeScript stack' for their tech: - Node.JS w TypeScript on the backend - React with TypeScript on the frontend - Some standard enough API approach (eg GraphQL) This is both easy to hire for + easy to pick up for any developer.
English
116
204
2.7K
0
Casey Chow retweetledi
Steve Krenzel
Steve Krenzel@stevekrenzel·
With Twitter's change in ownership last week, I'm probably in the clear to talk about the most unethical thing I was asked to build while working at Twitter. 🧵
English
1.2K
26.9K
127.9K
0
Casey Chow retweetledi
Roy E. Bahat
Roy E. Bahat@roybahat·
And to be fair, doing VC on easy mode (i.e., without fighting to be great) is *really* easy. Opine in some partner meetings, host an LP for a lunch and riff about the valuation weather, and hope nobody finds you out.
English
3
10
196
0
JJ Tang (Rootly)
JJ Tang (Rootly)@jjrichardtang·
Is it just me or does @opalelectronics never work with Google Meet or Loom? Only the preview in the native app seems to load up the video feed.
English
2
0
0
0
Casey Chow
Casey Chow@chasey_cow·
@mscccc There's a story in Domain Driven Design that I think is a pretty apt analogue to this.
Casey Chow tweet media
English
1
0
3
0
Casey Chow
Casey Chow@chasey_cow·
Over the past two days I've been getting texts from someone that I need to finish landscaping their lawn. If this is you, 1. I forgot all of my Chinese 101 already I literally only took that shit to graduate 2. nice house
Casey Chow tweet media
English
0
0
2
0
Casey Chow
Casey Chow@chasey_cow·
@sidpanj It’s actually a tactic—by making the message less believable on close inspection they weed out people who aren’t easy targets.
English
0
0
2
0
Sid Panjwani
Sid Panjwani@sidpanj·
They’re not even trying
Sid Panjwani tweet media
English
1
0
1
0