Joined May 2026
1 Photos and videos
Anyone has had success with open source models into production workflows?
1
Piyush Kumar retweeted
Replying to @crystalkemal
@crystalkemal is AWESOME btw!
We recently blew past the 5,000 mark and are now hitting 8,600 bi-weekly downloads across all Kemal @crystalkemal projects! 🤯 That’s an average of 610 downloads a day 📈 Building high-performance backends has never been this seamless. The raw speed of Crystal combined with the simplicity of Kemal is truly a game-changer. Next target: 1,000 daily downloads! 🚀 @CrystalLanguage #crystallang
1
3
6
293
Piyush Kumar retweeted
Replying to @AnthropicAI
In what universe does this not read like a manifesto for regulatory capture and suppressing competition? ANTITRUST. ANTITRUST. ANTITRUST.
4
186
4,267
Piyush Kumar retweeted
I don’t think my co-founder is particularly thrilled that I burned through 40B tokens last month 😅
I made a VS Code extension for when you are done token-maxxing. Spin of Shame: a tiny fidget spinner with cursed quotes, fake diagnostics, token sparks, and reverse inference mode. Does it help? No. But it spins. github.com/pi-erzan/spin-of-…
1
2
63
Piyush Kumar retweeted
I've been on the edge for months now but that's it! I'm DONE!
3
5
54
2,506
Heya #founders & #builders what are you building this week? How many users you’ve got so far? How did you get your first design partner/paying customer? Let’s #connect and share some #growth ideas!
3
20
I made a VS Code extension for when you are done token-maxxing. Spin of Shame: a tiny fidget spinner with cursed quotes, fake diagnostics, token sparks, and reverse inference mode. Does it help? No. But it spins. github.com/pi-erzan/spin-of-…
1
3
73
try it out and let me know if the spinner needs more torque
2
9
Most companies are gambling when making AI Agent changes. Its the new site-redesign, hoping it improves some vague metric
2
3
Piyush Kumar retweeted
Most “AI agent infrastructure” companies are secretly just building serverless platforms with better branding. The workloads agents need and the workloads serverless was designed for are fundamentally different.
1
2
2
85
Piyush Kumar retweeted
Looking to deploy durable, long-lasting AI agents. We would love to talk! We’re building Zunesha around this exact problem space: durable Linux machines for long running AI agents. Link in thread 👇
1
2
2
67
Piyush Kumar retweeted
Replying to @Guronnimo @X
Hey 👋 building the runtime layer for AI agents here at Zunesha Labs. If your stack runs agents, we should talk. 🎁 Join the waitlist and get $100 free compute credit → zunesha.ai
1
2
9