Sir Hakase (@codehakase) profile picture

swe: infra, distributed systems, ∀Go ∧ ∃!Haskell, 🎹 noob. Views are mine!

undefined
Joined July 2015
Photo by Sir Hakase (@codehakase)Photo by Sir Hakase (@codehakase)Photo by Sir Hakase (@codehakase)Photo by Sir Hakase (@codehakase)Photo by Sir Hakase (@codehakase)Photo by Sir Hakase (@codehakase)Photo by Sir Hakase (@codehakase)Photo by Sir Hakase (@codehakase)Photo by Sir Hakase (@codehakase)
A quick demo of floating panes in Rex, some of their features, and how I use them in my daily workflow. I think our implementation of floating panes looks great and feels great and I think you'll love it! These are also all scriptable and API driven, too, of course!
153
86
51
2,852
238,921
Sir Hakase retweeted
Thanks to AGI advances, Jobstack is now completely free. lighter models in use now and cheaper. total AI tokens cost today is <= $40/m now. Does not seem like a great idea to charge for that :)) jobstack.me
18
65
3
201
12,325
our S3 bill was way too expensive… so we built a 30 PB storage cluster in the heart of SF.
8
21
11
203
46,275
Sir Hakase retweeted
Your customers have questions. The answers already exist in your documents but people can’t find them fast enough when it matters. Pernoex adds a simple assistant to your site so users can ask and get help instantly. pernoex.com
2
25
6
31
9,072
Sir Hakase retweeted
We launched @byInfraRed's document fraud detection product last week 🚀 If you're a bank, digital lender, an insurance company, or a fintech that needs to be sure you're not accepting tampered documents, we built InfraRed for you Let us work with you: cal.com/infrared
1
55
58
4,586
Sir Hakase retweeted
Join the wailist for @useallo plzzzzz useallo.com
5
32
2
54
6,887
Let's talk about scrollbars in Ghostty. Didn't have time to blog, so decided to record a ~10 minute video describing the background on this problem and walk through how I modified the core of Ghostty to support scrollbars (coming soon!) If you like this content let me know.
24
21
4
793
50,517
Sir Hakase retweeted
After almost 2 years of building (before open-sourcing it), I’m excited to share Servflow, an engine and dashboard built to make API and AI agent creation ridiculously easy. What makes it special: ⚙️ Create AI Agents instantly, exposed over an API 🌐 Build standard REST API endpoints with rich integrations & workflow actions 🤖 Spin up full MCP servers powered by workflows 🧩 All declarative, all modular , designed for both power and simplicity 💻 Can be locally hosted and run Built with love using #golang Servflow is now open source, built to help developers move faster and prototype smarter. Check it out, break it, improve it, go crazy 👉 @servflowio
7
78
3
247
60,978
Minor medical situation on the flight and it’s cool that my wife is able to jump up and help out when they ask for a licensed medical professional. One day someone will need a regular expression so I stay ready.
112
213
25
5,148
163,381
Vitess databases on PlanetScale support sophisticated vector search. We've published a deep-dive by @vmg on how we built this for MySQL in a way that is scalable, performant, and ACID-compliant.
1
14
5
163
55,580
Sir Hakase retweeted
On my path to becoming a football coach, I saw the same problem everywhere: Match analysis takes too long, it’s slow and repetitive. That’s why I’m building Reyda, an AI-powered football analysis tool that turns hours of manual tagging and reporting into minutes.
26
159
31
272
33,162
We made improvements deploying static sites on @brimblehq. An average deployment time of about 4 seconds More improvements ahead 🫡 Signup here (brimble.io) to get started. @theresidency @nick_linck
1
12
83
2,154
Sir Hakase retweeted
WaitGroup.Go in Go 1.25 is indeed a quality-of-life change. I’ve seen (and made) the mistake of adding wg.Add(1) inside a goroutine, now that’s finally gone.
13
87
8
682
48,436
Sir Hakase retweeted
Amp Tab is off by default in Amp. More and more people are finding it and enabling it. New features like this need to earn their usage before we consider making them on by default, rather than riding the popularity of the other stuff. Try it out! I use and love it.
Amp Tab is now 30% faster We worked with Baseten to switch to TensorRT-LLM and KV caching with speculative decoding. The new lookahead decoding also reduces draft tokens and compute per iteration by using improved n-gram candidate selection and variable-length speculations.
4
9
1
63
8,575