SWE @ClearGlass_ | Former. GroupAlike App | Fullstack Engineer
Mumbai,India
Joined April 2017
- Tweets653
- Following490
- Followers130
- Likes1.6K
Pinned Tweet
Introducing Formacer: Your personal AI form assistant 🤖 . Save your information once, fill any form with a single click. Whether it's travel forms, job applications, or registration forms, Formacer has your back.
Download: chromewebstore.google.com/de…
Thanks for helping with this @pushpak1300, I was literally stuck at the data aggregation step!.
Is your paneer really paneer?
Can mangoes be ripened with calcium carbide?
Must a pizza chain show calories?
The law answers all three. Nobody reads it.
So I built fss .ai. Every answer is grounded in the law and links to the official gazette PDF.
fss.ai.dumbbellcode.in
Is your paneer really paneer?
Can mangoes be ripened with calcium carbide?
Must a pizza chain show calories?
The law answers all three. Nobody reads it.
So I built fss .ai. Every answer is grounded in the law and links to the official gazette PDF.
fss.ai.dumbbellcode.in
Went from 20ft to 14100ft above sea level.
5 hour climb, body says no, mind says let's go!.
Low on oxygen, high on adrenaline.
#HamptaPass
Sudheer Tripathi retweeted
100 MCP tools.
your agent does NOT need to meet all of them at once 😭
Laravel MCP v1 introduces searchable tool catalogs, keeping tools/list basically constant whether you have 10 tools or 100.
wrote about the whole rabbit hole here ↓
laravel.com/blog/a-better-wa…
Have you noticed working with claude code gets better as you use it in a codebase for long time.
The reason is claude code persists frequently occurring requests in it's own memory, as markdown files.
You can look at those files under /memory -> open auto-memory folder option.
This is true for code reviews too!
The MR author knows the intent of the change, the reviewer understands the intent, but the LLM that does the review doesn't. So it starts pointing out things in every nook and corner ! chasing perfection.
The harshest realisation for me after we’ve got LLMs is that there is no such thing as “good code”. Everybody’s code sucks. You can always take seemly good code and find bugs and improve the design, and this process doesn’t seem to converge. The goal I was chasing my whole career turned out to be a marriage: the closer you get, the further it appears. But LLMs allowed me to prove this empirically, so now I’m more at peace with my code. If it works after sufficient testing, I leave it alone; not because it’s “good” but because it’s good enough.
P.S. I ran Astra on this famous John Carmack’s code and it found five verifiable bugs.
Sudheer Tripathi retweeted
Completed one year at @laravel.
Here are my learnings 👇
What's the benefit of interpreted languages if you eventually end up in a jungle of linter, formatter, type checker and static analysis tools. With the tools themselves being slow, and then being eventually rewritten in compiled languages.
Compiled languages 🫡
Sudheer Tripathi retweeted
Cost per Intelligence Index, updated with the latest models.
For daily work I’ve always run Opus & GPTs on Medium, bumping to High for complex tasks and heavy iterative dev. The models clustered in the center are legitimately excellent daily drivers.
Grok 4.5 has been outstanding for me, feels right in line with GPT 5.6 Sol Medium (exactly as the chart shows). Only complaint: Grok please give us more resets like OpenAI. Once they kill the generous resets, surviving on weekly limits (no 5hr limits) is going to be tuff 😅
#ChatGPT #OpenAI #Sol #Grok #Composer #Opus #Fable #artificialanalysis #Anthropic
