Andrea Giammarchi 🍥 @WebReflection

Web, Mobile, IoT: all JS things since 00's - opinions are my own ™ Principal SW Engineer @anacondainc working on OSS stack 🦄 medium.com/@WebReflection Milan, Italy Joined May 2009

Tweets

45K
Followers

10K
Following

1K
Likes

16K

Andrea Giammarchi 🍥 @WebReflection

9 hours ago

@matteocollina Has anyone else answered DeepSeek already?

0 0 1 337 0

View Details

Mia @MiaAI_lab

14 hours ago

DeepSeek V4.1 Flash 1017 on 2x DGX Sparks Frontier at home. No limits. 100% Private. github.com/MiaAI-Lab/Deep…

Ray Fernando @RayFernando1337

22 hours ago

How is DeepSeek v4.1 Flash this fast on 2 DGX Sparks??? This is a national security risk.

127 84 3K 227K 749

29 33 580 34K 311

View Details

🚀 Introducing DeepInfer, our in-house inference framework, now fully open source! Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable. 🔗 github.com/deepseek-ai/De… 1/4

1K 4K 33K 4.0M 13K

View Details

Andrea Giammarchi 🍥 @WebReflection

21 hours ago

@thekitze DejaVuln as fallback 🤣

0 0 0 370 0

View Details

Andrew Curran @AndrewCurran_

a day ago

The Department of Homeland Security is requiring all civilian federal agencies to remove DeepSeek products and 'any third-party services developed using DeepSeek implementations,' and to stop buying them.

242 362 2K 555K 730

View Details

Andrea Giammarchi 🍥 @WebReflection

22 hours ago

@srcbuild that plan is called “sprint” 😉

0 0 0 10 0

View Details

Andrea Giammarchi 🍥 @WebReflection

2 days ago

underrated skill in this AI era: the scrum master! * definition of done (any AI loop) * split bigger tasks into incremental milestones (context compaction when needed) * story points per each task (energy consumption/deliverable ETA) missing: an automated AI driven kanban board

5 0 3 508 2

View Details

Andrea Giammarchi 🍥 @WebReflection

a day ago

@srcbuild “Not quite my tempo” … I have massive plan/goal and agents are working incrementally to adjust compaction, what are you talking about? 🤔

1 0 0 24 0

View Details

Mia @MiaAI_lab

2 days ago

Here's how 4 concurrent streams look on DeepSeek V4.1 Flash 1017 running on 2x DGX Sparks.

Mia @MiaAI_lab

2 days ago

Run DeepSeek V4.1 Flash 1017 on 2x DGX Sparks ✨ - 1M total KV pool, 512K context - 75 tok/s on prose, single stream - 130 tok/s on prose, 4 concurrent streams - Prefill 2000-2175 tok/s - EXL3-EAQv1, 2.9 bpw The quant in this recipe got about the same calibration as my GLM 5.3

78 62 745 82K 567

30 11 289 15K 59

View Details

Andrea Giammarchi 🍥 @WebReflection

2 days ago

about the "missing" bit, compaction and STATUS.md theoretically solve that but it's nowhere visually easy to follow on a timeline based board ... anyone that already solved this problem out there? thanks in advance for pointers 👋

0 0 0 102 0

View Details

Andrea Giammarchi 🍥 @WebReflection

2 days ago

TFW your local AI uses Python because it's convenient around math and then it writes all over: "Python round() is banker's — use floor(x+0.5)" 😅 ... what I love about this approach is that I am both learning but also realizing AI writes way more tests than I could imagine 🤯

0 0 0 338 0

View Details

Andrea Giammarchi 🍥 @WebReflection

2 days ago

@AdamRackis @dhh I think golden sponsors gave him either money or million of tokens so that both mean he has a lot more tokens than average one. Considering how far local AI went though, it feels weird professional developers are still worried about their tokens consumption, imho

0 0 1 144 0

View Details

Andrea Giammarchi 🍥 @WebReflection

3 days ago

@MiaAI_lab TFW your only OC agent is local so you feel for others and yet care not nearly enough for yourself 😅

0 0 0 924 0

View Details

OpenCode @opencode

3 days ago

We had a security breach last night related to a large scale distillation attack that impacted 1.4% of OpenCode Console users We have notified them all this morning via email. If you did not get an email you were not impacted, it was a subset of users who had external providers linked The incident is ongoing so we cannot share more details at this point but we are working with Anthropic, OpenAI, and Google to remediate

725 1K 12K 2.9M 2K

View Details

Andrea Giammarchi 🍥 @WebReflection

3 days ago

if context is needed, around flatted, filing issues around it not being JSON compatible is literally written in the README it’s not: it’s like failing issues to PNG je cause it cannot decode JPEG which is an image … I don’t have time for this, I’m afraid

0 0 0 88 0

View Details

Andrea Giammarchi 🍥 @WebReflection

3 days ago

is @GooglePlay reasonable? "Submit a polished, bug-free app" ... sure, that's what everyone does on the very first version of any app, right? Gosh I've been told Apple time to market and requirements are much easier, I wonder why Google Play has to be so cumbersome these days.

0 0 0 394 0

View Details

Andrea Giammarchi 🍥 @WebReflection

4 days ago

@AdamRackis youtu.be/BPRTv9ckPEM

0 0 1 101 0

View Details

Andrea Giammarchi 🍥 @WebReflection

4 days ago

@azivc @StateOfJS still cooking … 🧑‍🍳

0 0 0 7 0

View Details

Andrea Giammarchi 🍥 @WebReflection

6 days ago

hopefully this will age well: while my agent is cooking, be assured Google Play and Apple Store are *done* feels like a clickbait Xweet, wait for my follow up post 👋