fde @realfastai | तत् त्वम् असि | building @spently

Delhi, IN
Joined March 2019
Best Cybersecurity Awareness Month ever. 😭
49
761
61
5,387
78,968
I was on Zerodha’s Subtext recently, talking to @pranavmanie and @bebhuvan about what an AI-native firm is, and the part people kept DMing me about afterwards was the few minutes on the middle tier of talent disappearing. This essay is the longer answer. sidu.in/essays/after-ai-ther…
7
30
6
231
15,776
the truthest truth i've heard
18
chill. i just work here.
403
1,474
631
111,381
10,699,758
True. All my projects, including @spently are completely based on @Cloudflare
migrated to cloudflare d1 for database r2 for storage $0/m vs $58/m thousands of users unlimited projects ts is 2026 meta
1
2
69
Ramanujan wasn’t skipping proofs. He was hiding his chain of thought to prevent distillation. Hardy: “Show your work.” Ramanujan: “Sorry, reasoning tokens aren’t exposed via the API.”
80
710
70
9,829
997,030
this does not look good at all @Apple after the golden gate update
1
31
this is the real dopamine when you work with good models
1
66
This is my new shortest “what is Jev” explanation
207
1,125
271
10,715
592,481
are there benchmarks that measures how much slop a model produces/how clearly it talks?
3
24
got jev access! thanks @typesafeai
atp not having either of jev/muse access is giving me fomo
1
25
atp not having either of jev/muse access is giving me fomo
1
1
61
loved the @typesafeai home page. art
1
18
px0 is amazing! - been using it for 2 days P.S. told codex to show me changes in px0 and it instantly understood what i meant and how to (sol btw, not even astra) @claudeai needed explanation as to what it is.
Agentic inline editing coming soon in px0.ai :) Also, px0 hit 1k stars on GitHub in less than 3 days. Thank you so much for all the love and support. Thank you for all the PRs and issues. Keep them coming. So many genuinely good features and bugs; working overtime to merge and squash them. Loving it... Keep using px0 and keep spreading the good word about it. The idea is to build a fast (< 1 ms), ultra-light (~20MB mem), remote-first IDE for agentic SDLC. Star the repo if you find px0 useful. It would mean the world to me. github.com/px0-ai/px0
1
3
287
i shed a tear reading this blog post by a deepseek kernel engineer
64
463
61
6,730
211,373
Avi retweeted
you seek more intelligence, Arjuna, when what you lack is clarity
31
479
28
5,358
76,998
have basically shifted completely to codex now
1
22
sessions running for months, contexts bloated with hundreds of skills and sessions plauged with compactions - the sorry state of affairs in the workbench of ai-natives (or, just me being an old man yelling at claudes)
25
21
8
168
17,507
All-timer from Codex.
49
600
39
18,406
757,587
my weekends look like this now
2
44