life.exe still compiling...
india
Joined October 2024
- Tweets865
- Following324
- Followers157
- Likes3.3K
Shivam retweeted
We are entering a new era of discovery now:
openai.com/index/sharing-ai-…
Shivam retweeted
How important is AI to Meta? Here’s one easy tell:
One of the launch features of the Muse app was auto-negotiating for stuff on Facebook Marketplace.
The poor product manager who owns Marketplace knows that millions of bots lowballing sellers will eventually kill the product.
But this was escalated to Zuck by the Muse team and he said: “OK, we’ll poison Marketplace…but just a little.”
Shivam retweeted
I was about to pay AWS for storage.
Then I remembered my Google Drive has 2TB sitting empty.
So I built an API that turns your Drive into an S3 bucket.
Upload, download, signed URLs, versioning — all through a normal REST API.
Your storage. Your Drive. Zero cloud bill.
PH:- producthunt.com/products/byo…
Shivam retweeted
Give it a read: wsj.com/opinion/the-ai-futur…
Shivam retweeted
Ended up training an RL agent with Soft Actor-Critic in a custom Gymnasium environment to handle execution instead of TWAP/VWAP schedules.
Beats TWAP by ~35% and VWAP by ~30% on implementation shortfall (2.71 bps vs 4.18 bps).
Live at execagent.vercel.app/
Shivam retweeted
If you want to start a startup, don't learn "entrepreneurship." Learn how to build things. The hard part of startups is not "entrepreneurship" but product: to know what to build, and to be able to build it.
Shivam retweeted
insult me. call me a vibe coder, i don't care. i'm the future of this industry, not you. yeah i don't know anything about time complexity, how a websocket works, and i still don't know if hash map is a real "data structure" (?) or not. but my ai code is in your favorite open source project. maybe it improved it, maybe it broke it, but it's there. when's the last time you gave a shit about any of that anyways? my github graph is green green green baby. why cant you, the "real swe" keep up? "oh no thats all ai slop jimmy you aren't a real engineer" man shut up. 4 years ago a lot of you were landing 1 pr a week while spending most of your time arguing about vim vs vscode or how many monitors you need to be a "cracked" swe. that was the real slop, not my 1 million line diff i want you to merge. maybe my code exposes user information or racks up a million dollar vercel bill but thats what moving fast means. if you care so much about the boring shit then why don't you fix it while i innovate. so go ahead and laugh. keep saying i'll never be a real engineer. i'm already in the repo.
Shivam retweeted
Many people think any given ML project is 99% training.
In reality, it’s 50% evaluation, 40% data cleaning, 8% integration, and 2% training.
The first two set the noise floor for learning. No ML magic matters; the model cannot lower the noise floor, as that’s the optimal bound of Shannon encoding of your data.
Thus, not a single day goes by without me thinking about ontology. Even the old labels have to be constantly reviewed.
