Sohan (@HiSohan) header
Sohan (@HiSohan) profile picture

Sohan

@HiSohan

I collect dots to connect them later. Views are my own. Building https://nitter.cf/t.co/QIx831U6wn Writing on https://nitter.cf/t.co/rOzI5yJv5t

 Bengaluru, India
hisohan.com
Joined December 2011
  • Tweets7.2K
  • Following621
  • Followers1.5K
795Media
Photo by Sohan (@HiSohan)Photo by Sohan (@HiSohan)Photo by Sohan (@HiSohan)Photo by Sohan (@HiSohan)Photo by Sohan (@HiSohan)Photo by Sohan (@HiSohan)Photo by Sohan (@HiSohan)Photo by Sohan (@HiSohan)Photo by Sohan (@HiSohan)Photo by Sohan (@HiSohan)
The philosophies around learning between Tagore and Socrates look very different. Tagorean Curiosity is one of wonder and bringing a fresh perspective. Socratic Inquiry is based on doubt - to reveal a deeper truth by questioning I love both.
Image from Sohan (@HiSohan)
3
1
2
17
6,756
Such a classic power move! – at Bongaigaon Refinery, India
Image from Sohan (@HiSohan)
1
75
iPhone trio. – at Bongaigaon, India
80
I don't walk back this, but yes, reading and reviewing every line of code is perhaps less important.
Sohan (@HiSohan) profile pictureSohan
@HiSohanJul 24
I was never a fan of unc. I always hated clean code and this is rather hypocritical to come from him. He used to fearmonger saying if you don't write clean code, something will fuck up and regulators will get involved and start issuing licenses for SWEs. Not reading code risks exactly that. That means it was always about selling books and very rarely about building an engineering culture that values practicality, durability, reliability and versatility.
50
This is extremely clever, I am not sure how useful it is, but I can see some utility.
div (fp32) (@divagr1925) profile picturediv (fp32)
@divagr1925Oct 28
1/4 Meet Qwenputer: Qwen3.5-4B with a full RISC-V CPU inside its weights, using the model's own Gated DeltaNet layers as RAM. It runs C and MicroPython in its forward passes, no external runtime. On DeepMind math: base 55–90%, base + Python tool 95–100%, Qwenputer 97.5–100%.
1
80
This is unequivocally, 100% true.
hrishikesh kamath (@recursiveagi) profile picturehrishikesh kamath
@recursiveagiOct 29
The first problem with AI-generated text detectors is the assumption that 100% AI-generated means zero human thought. A non-technical person and @karpathy could both generate 100% of their code using AI. That doesn’t mean the code they produce reflects the same level of engineering skill. The same applies to writing.
1
1
100
Sohan retweeted
Paul Graham (@paulg) profile picture
Paul Graham
@paulgOct 28
17 yo never listens to us and yet hears everything we say about him, even in a whisper. I think he runs a side process using 2% of his brain listening for his name, and if he hears it, then he actually interprets the phonemes he's been buffering.
182
13
12
1,300
150,625
If Paras runs the ministry of compute, I can more or less guarantee that effective compute per capita is immense.
Paras Chopra (@paraschopra) profile pictureParas Chopra
@paraschopraOct 28
India needs a ministry of compute. Like ministry of oil, we need laser-focus from the top to build compute necessary for participating in the future economy. Otherwise, we will repeat in a decade what we’re doing now: spending expensive forex for importing what we don’t have.
5
31
1,025
Will be*
53
Kickstarted Batch 9 of @lossfunk 🚀 residency! We took a short pause but now we are back with a fresh set of curious folks and deep projects. This batch has it all: AI agent swarms, cancer drugs, AI sitcoms, a GTA-style Bangalore, and LLMs served directly from OS. 🧵 with their profiles and goals for the next 6-8 weeks.
Image from Paras Chopra (@paraschopra)
41
23
15
547
48,177
you either find a PMF or live long enough to become an RL environment company
4
1
18
699
Get a helmet, anon! @reddy2go – at Bongaigaon, India
Image from Sohan (@HiSohan)
2
5
147
some here be like
Image from 𝖌𝖎𝖘𝖙 (𖡎 era) (@lfgist)
123
684
75
15,035
487,808
Nothing beats hard work. Not even smart work. Privileged people who get lucky and haven't self actualized spread this whole smart work beats hard work rumor. Hard work builds smarts too. You can't become smart by being lazy. – at Bongaigaon, India
1
3
74
Not sure if it's satire or a prophecy.
Luckey Faraday (@luckeyfaraday) profile pictureLuckey Faraday
@luckeyfaradayOct 25
If your software is human-coded slop, I’m not using it. It’s just too risky. AI can find vulnerabilities in human code faster than humans can find and patch them. I’m done trusting software written by humans. AI-coded only from now on.
53
Ever since the tesla moment. I always say, today, we have cars that fly. What's your excuse and it never gets old. – at Bongaigaon, India
1
87
Last month, Chinese researchers demonstrated quantum teleportation over 100 simultaneous channels. Who's building a verifier for quantum computations? 🤔
Image from Nilesh Trivedi (@nileshtrivedi)
Image from Nilesh Trivedi (@nileshtrivedi)
2
8
703
Some techbro wrote that humans will never be as good as LLMs. Bro used to sell coding courses.. My problem is not the claim, but rather a lack of evidence. I too think that LLM aided development is the future, but there's nuance to it. – at Bongaigaon, India
3
238
Engineers and technical people have very different prompting styles. – at Bongaigaon, India
1
51
LARPing with chatgpt is insanely addictive.
3
89
Please implement this.
Paras Chopra (@paraschopra) profile pictureParas Chopra
@paraschopraOct 23
Strategic laziness: waiting on difficult projects because a future AI will be able to one-shot it much before you would have completed the project yourself.
69
Load more