Sriram Sivakumar

88 posts

Sriram Sivakumar

Sriram Sivakumar

@hashsriram

AI infra & agentic systems. Running production-grade local LLMs on a single Grace Blackwell. Sharing what works (and what doesn't). Singapore 🇸🇬.

شامل ہوئے Aralık 2025
419 فالونگ81 فالوورز
mr-r0b0t
mr-r0b0t@mr_r0b0t·
@hashsriram @NVIDIAAI Yes! #3 is where I will wind up for a while is think! Hopefully by the time I can afford 4 there will be a GB20 I can hook these up to 🤣
English
1
0
1
52
mr-r0b0t
mr-r0b0t@mr_r0b0t·
So I did a thing 😁
mr-r0b0t tweet media
English
31
0
98
4.3K
Sriram Sivakumar ری ٹویٹ کیا
Pavel Durov
Pavel Durov@durov·
We now support rich formatting for all chatbots. Tables, nested lists, inline media, formulas, headers and more — right in Telegram messages. 🔨 Start building! Docs: #rich-message-formatting-options" target="_blank" rel="nofollow noopener">core.telegram.org/bots/api#rich-…
English
328
332
5.4K
297.3K
Luke@DeskPi
Luke@DeskPi@Mandy_Li888·
@hashsriram @NVIDIAAI I spotted two DeskPi RackMate 10-inch 2U Dual Cooling Fan Panels 👀 Now I’m really curious what hardware is hiding inside that rack — mini PCs, SBCs, storage, or something even cooler?
English
1
0
0
81
Sriram Sivakumar
Sriram Sivakumar@hashsriram·
@the2ndfloorguy check this out buddy! Took your light bulb idea and implemented into this system. Loving it so far!
English
0
0
0
56
Andrew
Andrew@andrew_Tas·
@hashsriram @NVIDIAAI Love it! how did you attach the ipad to the rack? i hate double sided tape.. haha
English
1
0
1
193
Pankaj
Pankaj@the2ndfloorguy·
My zerodha stock portfolio now controls my bedroom lights. It turns red when I lose money.💸 🔻
Pankaj tweet mediaPankaj tweet media
English
359
309
9.5K
806.5K
mr-r0b0t
mr-r0b0t@mr_r0b0t·
@hashsriram Do you like your bar monitor? Is that the Corsair one?
mr-r0b0t tweet media
English
1
0
0
33
Sriram Sivakumar ری ٹویٹ کیا
Sundar Pichai
Sundar Pichai@sundarpichai·
DiffusionGemma is an open, experimental model that brings our text diffusion research to Gemma 4. It’s a racehorse 🏇achieving up to 4x faster inference by generating entire blocks of text simultaneously vs predicting token-by-token (word-by-word) output!
English
175
396
3.3K
292.2K