
Jithu R Jacob
@jithu_r_jacob
Tech Lead, ML Engineer. I tweet about models, systems, products, prod deployments & engineering. Opinions are mine and not that of my employer.
你可能会喜欢
Market research firms are cooked 😳 PyMC Labs + Colgate just published something wild. They got GPT-4o and Gemini to predict purchase intent at 90% reliability compared to actual human surveys. Zero focus groups. No survey panels. Just prompting. The method is called Semantic…

Did you know you can run ngrok (without installing it) by using tunnel.us.ngrok.com as a reverse tunnel? Try this: ssh -R 80:localhost:80 http It's the equivalent of `ngrok http 80`

This paper shows that you can predict actual purchase intent (90% accuracy) by asking an LLM to impersonate a customer with a demographic profile, giving it a product & having it give its impressions, which another AI rates. No fine-tuning or training & beats classic ML methods.



Unpopular opinion: Software engineering is the only job where the interview is harder than the actual job Interview: Design a distributed system for 10M users Job: The login button moved 2 pixels, please fix
Server load so high you can’t even SSH in? 👉 Use this command; ssh -o ConnectTimeout=1 -o ConnectionAttempts=1 user@host "nice -n -20 bash" It gets you a high-priority shell when nothing else works.
💡 DB Design Pro-tip: Soft Deletes Soft deletes: instead of removing a row, you mark it as deleted (e.g. deleted_at TIMESTAMP). Great for undo, audit trails & not breaking FKs. But! If you query the base table directly, half your code will forget to exclude “deleted” rows.…
🚨Terrifying news in TOI today – 1 in 3 Indian children aged 5–9 already have high triglycerides, and Delhi tops in hypertension. Why Are Indian Children (5–9 yrs) Showing High Triglycerides? The report shows that 1 in 3 Indian children aged 5–9 already have high triglycerides,…

Run 70B LLMs on a 4GB GPU with layer-wise inference and memory optimization, quantization optional

A lot of opportunities here for Indian hardware startups.

Tip: You don't need python-dotenv, use "uv run" with --env-file, and your env vars from .env get loaded. For example, if you've got a @FastAPI project you can run it locally with env vars like: uv run --env-file .env fastapi dev
"You need to make yourself a big target for luck, and the way to do that is to be curious." Paul Graham on how to figure out what to work on:

Almost tearing up seeing this! Government of India is working towards standardizing the UX across its services and has even a Figma design system out.

Why do people call workflows "AI Agents"?

How do we measure AI fluency at Zapier? Here are some role-by-role examples 🧵

We’re setting a new standard at Zapier. 100% of new hires must be fluent in AI.
I am advising my obese liver disease patients aerobic exercise and strength training, and to go on a calorie deficit diet to lose weight, but everyone is sitting home eating chia seeds four times daily and drinking green tea twice a day with lemon & honey shots on empty stomach.…
Since I wrote this answer to a student, I thought I would share more broadly.

Large Language Models don't reason. Thank you, Apple.

Uber's scale is a master class example of what a world-class data infrastructure stack looks like. 👌 A must see for any data engineer. More info available tomorrow for all 2minutestreaming subscribers. Not subscribed? Join 5000+ readers by checking the letter in my profile.

United States 趋势
- 1. #LoveYourW2025 96.7K posts
- 2. Good Wednesday 20.3K posts
- 3. #wednesdaymotivation 4,698 posts
- 4. #Worlds2025 46.7K posts
- 5. Hump Day 7,381 posts
- 6. #LeeKnowXGucci 5,192 posts
- 7. Raila 213K posts
- 8. And the Word 74.8K posts
- 9. LEE KNOW FOR HARPERS BAZAAR 4,589 posts
- 10. #Wordle1579 N/A
- 11. Young Republicans 86.9K posts
- 12. Tami 4,828 posts
- 13. Baba 121K posts
- 14. Yamamoto 51K posts
- 15. Lucia 57.4K posts
- 16. Vishnu 9,046 posts
- 17. halsey 9,957 posts
- 18. George Floyd 36.5K posts
- 19. Politico 325K posts
- 20. Burl Ives 1,792 posts
Something went wrong.
Something went wrong.