Building @typefully | indieHacker, shopify dev, public speaker

Your internet
Joined April 2010
You all must have heard people saying "build something that solves your own problems" Here is how building an app to solve my own problem did wonders for me 👇
7
12
8
138
the strongest of eMudhra's 8 keys was 237, a 3-digit luggage lock has more combinations
74
Would like to see how much of that loss is training models vs serving models. Training at a loss is something I can digest, serving at a loss is a lot scarier
JUST IN: Anthropic recorded a staggering $80,000,000,000.00 net loss in the first half of 2026.
2
76
Rajat Kapoor retweeted
The beautiful thing about the PHP bug is that all it takes is one dedicated hero to use the CVE to defeat the CVE (mitigate public facing PHP systems on their owner's behalf)
1
3
27
1,579
Tesla fanbois and Musk had hyped the Roadster demo so much that showing a very good car might have actually disappointed people. The only thing that could have saved them was a flying car...
4
186
Humble request to the frontier labs to please exclude this out of their training data
Howdy partner hee haw at 91 at 1 having one, baap baap hota hai, he is the real monster, hum sab ke baap mein hai dum Vande Mataram. Ek aur din nahi hota ek aur din kam hota hai, u guys get it? God’s been kind ..
79
in Bangalore 38 minutes barely gets you to the end of your street
There is no faster city for flying than San Francisco. 9:10pm: Leave house 9:32pm: Arrive at SFO 9:40pm: Get to gate 9:43pm: Boarding starts 9:48pm: At seat I’ve never done a 38min house-seat time before.
1
7
379
Rajat Kapoor retweeted
AI existed before LLMs got mainstream. I fail to understand why these companies are just doing this now and did not train NNs earlier with the data that they already had
We were able to fine tune an open weight model (Qwen3.5-9B) on proprietary internal data at @coinbase, and it out performed frontier models, at much lower cost and speed, for a specialized task (fraud prevention in this case). I think we'll see more of this pattern inside companies. The frontier models are amazing generalists. But increasingly you'll see specialist models out perform on more narrow tasks, either working on teams, or by themselves. We're making it easy to fine tune models internally for our employees. We even have a skill for it in our internal harness that makes it easy to spin these up and start training them: "/fine-tune-and-serve-llm I want to fine-tune Qwen for payments risk review. help me get started" I imagine we'll have some backend services that could be replaced entirely by this new architecture.
1
2
131
Rajat Kapoor retweeted
.ai is Anguilla's country domain and .si is Slovenia's So the .ai to .si domain switches quietly moves domain money from one tiny country to another
1
1
1
1
153
Rajat Kapoor retweeted
The GPU duopoly is one family WhatsApp group
The guy who runs Nvidia and the woman who runs AMD are cousins. Today the President handed them both the National Medal of Science on the same stage. Jensen Huang's mother is the younger sister of Lisa Su's grandfather. Their family patriarch in Taiwan had 12 children, and two branches of that tree ended up running the two companies that make the chips the entire AI boom runs on. They didn't even grow up together. Jensen's parents sold nearly everything they owned to send him to boarding school in rural Kentucky at age 9. Su's family brought her to America at age 3. The cousins met for the first time as adults, at an industry event, decades into their careers. Jensen's first chip job was actually at AMD. He designed microprocessors there before founding Nvidia in 1993. Su took over AMD in 2014 when the stock traded around $3 and spent the next decade reviving it with chips built to compete directly against her cousin's. And zoom out on the full list from today. Sergey Brin was born in Moscow. Satya Nadella in Hyderabad. Elon in Pretoria. Jensen and Lisa in Taiwan. Five of the six people receiving America's highest science and technology honors started life somewhere else. One grandfather in Taiwan had 12 kids. Two of his descendants now hold the National Medal of Science, earned at companies built to beat each other.
1
1
1
348
.ai is Anguilla's country domain and .si is Slovenia's So the .ai to .si domain switches quietly moves domain money from one tiny country to another
1
1
1
1
153
AI existed before LLMs got mainstream. I fail to understand why these companies are just doing this now and did not train NNs earlier with the data that they already had
We were able to fine tune an open weight model (Qwen3.5-9B) on proprietary internal data at @coinbase, and it out performed frontier models, at much lower cost and speed, for a specialized task (fraud prevention in this case). I think we'll see more of this pattern inside companies. The frontier models are amazing generalists. But increasingly you'll see specialist models out perform on more narrow tasks, either working on teams, or by themselves. We're making it easy to fine tune models internally for our employees. We even have a skill for it in our internal harness that makes it easy to spin these up and start training them: "/fine-tune-and-serve-llm I want to fine-tune Qwen for payments risk review. help me get started" I imagine we'll have some backend services that could be replaced entirely by this new architecture.
1
2
131
Rajat Kapoor retweeted
People are arguing whether it even matters whether we understand the proofs AI provides for the research it does. Programmers have already setlled on not reading code reviews line by line as long as the tests pass. But is that right for all math & science research? Thr debate has moved fast from LLMs failing at counting letters to arguing whether humans should care to understand their proofs
This is a slide from Terence Tao's 'Math 2.0' public lecture yesterday evening at Caltech. I am highly confident that, if the cure actually worked, almost any Stage 4 cancer patient would tell you that human comprehension of its mechanism is completely irrelevant to them.
1
1
114
People are arguing whether it even matters whether we understand the proofs AI provides for the research it does. Programmers have already setlled on not reading code reviews line by line as long as the tests pass. But is that right for all math & science research? Thr debate has moved fast from LLMs failing at counting letters to arguing whether humans should care to understand their proofs
This is a slide from Terence Tao's 'Math 2.0' public lecture yesterday evening at Caltech. I am highly confident that, if the cure actually worked, almost any Stage 4 cancer patient would tell you that human comprehension of its mechanism is completely irrelevant to them.
1
1
114
The GPU duopoly is one family WhatsApp group
The guy who runs Nvidia and the woman who runs AMD are cousins. Today the President handed them both the National Medal of Science on the same stage. Jensen Huang's mother is the younger sister of Lisa Su's grandfather. Their family patriarch in Taiwan had 12 children, and two branches of that tree ended up running the two companies that make the chips the entire AI boom runs on. They didn't even grow up together. Jensen's parents sold nearly everything they owned to send him to boarding school in rural Kentucky at age 9. Su's family brought her to America at age 3. The cousins met for the first time as adults, at an industry event, decades into their careers. Jensen's first chip job was actually at AMD. He designed microprocessors there before founding Nvidia in 1993. Su took over AMD in 2014 when the stock traded around $3 and spent the next decade reviving it with chips built to compete directly against her cousin's. And zoom out on the full list from today. Sergey Brin was born in Moscow. Satya Nadella in Hyderabad. Elon in Pretoria. Jensen and Lisa in Taiwan. Five of the six people receiving America's highest science and technology honors started life somewhere else. One grandfather in Taiwan had 12 kids. Two of his descendants now hold the National Medal of Science, earned at companies built to beat each other.
1
1
1
348
Every agentic shopping demo skips the part where an Indian card payment needs an OTP from your phone
3
170
Rajat Kapoor retweeted
harsh phrase im stealing from the internet meat proxy its an engineer who copies a task from slack into claude, waits, then pastes the pr link back into slack thats not a job if all you do is move a prompt from one window to another i can just do that myself the engineers who matter now are product-first or foundational everyone in the middle is optimizing themselves out of the loop
32
14
10
312
28,358
Rajat Kapoor retweeted
364
2,837
87
19,499
241,308
Are you sure you won't take it back tomorrow?
We just let in every Pro and Max user from the Claude Code Projects waitlist! If you're new to Claude Code Projects, here's a 4 minute walkthrough to get you started:
1
5
237
google discovering the best way to win the model race is to not need to win it
Google just announced that Gemini is not a model it's actually the agent harness, and they are going to add Claude models
107
What a sin! Now it's a race against time
2
140