in hiding
Joined December 2020
- Tweets 2,597
- Following 825
- Followers 7,028
- Likes 9,161
425 Photos and videos
Jun 12
a month ago, I asked a bunch of people I know why they switched to codex from Claude code and the tldr was that gpt 5.5 was just smarter
reasons people gave me:
-better able to understand problems and break them down
-“more autistic”
-auto compaction is very good
-oai is giving their team unlimited usage for free to test right now
-ability to use other harnesses (pi, opencode)
1
17
2,084
Jun 12
Twitter was generally unhappy with oai then, but it didn’t really affect developer choice behind the scenes. what ppl tweet and what they pay for are different
I tried it after some of these convos and have mostly been using codex since
1
5
933
Jun 12
I can’t help think the broader backlash against Anthropic is partially just downstream of vibes on model intelligence, people don’t realise fable is good yet
1
5
531
Nitya retweeted
Jun 10
Human biology matters. Scientists and AI need human data to understand health and disease.
Crownlands is open sourcing Gateway 4M, the largest single-cell tissue dataset ever released from living humans, to advance research on brain aging and neurodegeneration.
17
34
134
33,895
Jun 12
Murakami ran a club for 10 years and at the age of 30, started both writing full time and running every single day
this isn’t a book I particularly liked but it’s interesting how much running has clearly impacted his writing style
3
13
830
Jun 12
the recurring bit in every Murakami novel is that the protagonist goes on a grand adventure only to realise the thing or person they were looking for was themselves, or in front of them all along. running is a very similar 1:1 sport
5
184
Jun 6
the day I actually understand this book all of you will know
Jun 5
Warning: once you learn category theory, you'll never be able or willing to talk with people who don't know category theory.
7
55
955
71,232
Jun 6
May 4
“I will try to tell you what you need to know, and no more”
if you haven’t read Ravi vakils book on category theory do yourself a favor and read the first chapter so you know what good technical writing can look like
4
3,577
Jun 5
speaking at localhost! come listen to me talk at a conference about things I work on think about every day
. @nityasnotes from @ExaAILabs is joining us at Render localhost, June 18 in SF.
Web search for agents in production is harder than it looks. She'll walk through architecture, sub-agents, and what it takes to make it reliable at scale.
Save your spot: localhost.render.com
2
1
68
7,771
Jun 4
I've worked in biotech to AI and the main path to doing interesting work in new fields is that you have to take personal time to learn the fundamentals, there is no shortcut around it
this should feel difficult
6
2
74
4,970
Jun 4
approximate knowledge of many things can get you far but especially with AI, deep understanding of a few things gives you much more leverage to do things well
@justinskycak has a bunch of posts around this for math learning in general
1
9
635
Jun 4
I MISS NYC
My barber said let’s go spurs jokingly and I got out the chair and left. When will people understand. This is not a fucking game.
1
1
7
2,232
Jun 4
that tweet is how I want to react to things in my brain but I stop myself 101
1
970
Jun 2
this book feels like it was written exactly for my brain I'm taking a week to read each chapter
2
3
625
May 27
trying to work in 2026:
- slack is down
- github isn't working
- cloud provider blocked your account
software engineering is over
4
1
45
2,223