Building AI products in public - Indie Hacker - Content Creator - Ambassador @Torcdotdev 🚀 https://xitter.cf/t.co/BUqc64ardA
India
Joined May 2020
- Tweets9.3K
- Following1.1K
- Followers1.7K
- Likes23.6K
Arvind retweeted
🚨 A friend’s company is hiring in Bangalore!
They’re looking for:
Senior Product Manager
• 6+ years of experience
• Background in data & analytics, and related tools/platforms
• RGM experience preferred
Senior / Lead Data Scientist
• 5–8+ years of experience
• Master’s or PhD preferred
• Strong Python/R skills
• Experience deploying models in production
Both roles are Bangalore-based / office roles.
If you’re interested, drop your resume/details in comment or in my DMs or tag someone who might be a good fit. 🙌
🚨 A friend’s company is hiring in Bangalore!
They’re looking for:
Senior Product Manager
• 6+ years of experience
• Background in data & analytics, and related tools/platforms
• RGM experience preferred
Senior / Lead Data Scientist
• 5–8+ years of experience
• Master’s or PhD preferred
• Strong Python/R skills
• Experience deploying models in production
Both roles are Bangalore-based / office roles.
If you’re interested, drop your resume/details in comment or in my DMs or tag someone who might be a good fit. 🙌
Arvind retweeted
Jev is the "Internet" moment for AI.
200x faster and 400x cheaper than Claude Fable or GPT-6.
Here's 8 tips on how to master Jev:
- Stop treating it as an LLM. Jev doesn't write code.
- Study the three primitives: Choice, Score, Noul
- Use Choice when Jev should pick from a list.
- Use Score when you want a value on a scale.
- Use Noul for yes/no decisions with probability.
- Batch your questions. Jev can answer them all at once.
- Ask for richer output. Output tokens are free.
- Keep rules and math in code. Use Jev for judgment.
The architecture: LLM generates, Jev decides, code controls, humans handle uncertainty.
To learn how to use Jev, check out my latest article: craftbettersoftware.com/p/je…
Don't use expensive frontier models for decisions.
Use Jev to get 100x faster and cheaper results.
Arvind retweeted
Most common reasons for failing in order:
1) Never starting
2) Quitting too soon
3) Not doing enough volume
4) Not doing it consistently enough
5) Doing too many different things
6) Not shooting big enough
i just listed shotlyai.com on outliked, the leaderboard where you participate by posting and win based on likes.
every like on this tweet pushes it closer to #1.
do your thing 💗
outlike.lol/?via=shotlyai.co…
Stanford just released a free course on one of the most valuable skills in AI right now:
Building self-improving AI agents.
9 lectures covering:
• Verifiers & feedback loops
• Tools, code & memory
• Planning & reasoning
• Deep research agents
• Evaluating long-horizon tasks
• RL & test-time compute
Watch it. Build alongside it. Learn how agents improve themselves.
Full playlist 👇
youtube.com/playlist?list=PL…
House MD (my favorite one so far)
Supernatural
Breaking Bad
Stranger Things
Parks and Recreation
Brooklyn nine nine
Lucifer
Mr Robot
I tried running Claude Code loops for generating Three.js scenes, and my laptop completely froze. 😅
Next, I'm going to try the same workflow on my Hetzner VPS.
Curious, what setup are you using to run loops and subagents in parallel?
Today I learned about slang Jam juffered which is an internet-slang or playful malapropism/corruption of the phrase "shuffled as I have suffered"
I've been working on the same project with Claude Code for the past few days.
One feature that's saved me a lot of time: resuming previous conversations.
• Most recent:
claude -c (or claude --continue)
• Choose from past sessions:
claude -r (or claude --resume)
• Resume a specific session:
claude --resume <session-id>
No need to re-explain your project every time. 🚀
I really enjoyed and learned a lot from the sessions that you did in the past year.
Definitely recommend it to others 🚀
I’ve been busting my butt on @builtfordevs this year and it’s finally performing the way I want.
🚦Live signal that converts to a developer journey map you can measure and improve
🤖 Agent evaluations that act like your exact ICP so you know where your experience lacks. And if it’s agent-ready.
👨💻 Human evaluations that match your ICP using your dev tool and sharing their candid feedback along the way.
These three layers can now accurately tell you if your docs are lacking, is your marketing messaging isn’t hitting, or if your SDK throws an error.
🛠️ Recommendation engine gives you a list of things that you can fix to improve the experience. We’ll even do some of the recommendations for you if you want best-in-class docs or content improvements.
14-day free trial on the Build plan. Give it a try. It’s gold.
builtfor.dev
5 months left in 2026.
Challenge to myself:
🚀Ship 1 indie hacking project every month.
5 months.
5 launches.
Build. Ship. Learn. Repeat.
Holding myself accountable. Let's do this.
🤖 Made with AI
