Sandeep Gangarapu

1.8K posts

Sandeep Gangarapu

Sandeep Gangarapu

@sandeep_g_kumar

Applied Scientist at Big Tech. Ph.D. UMinn. A/B testing, Applied ML. Ex-Googler. IITKGP. Likes and RT's are notes to self.

Katılım Temmuz 2010
853 Takip Edilen547 Takipçiler
Sandeep Gangarapu
Sandeep Gangarapu@sandeep_g_kumar·
Everyone organizing Downloads folder using Claude cowork. Why? Who organizes Downloads folder?
English
0
1
0
60
Sandeep Gangarapu
Sandeep Gangarapu@sandeep_g_kumar·
anything that is a reference will got to apple notes. ex:nIf I browsing and I want to save it for later, I use this shortcut that uses AI @beardedteacher/video/7568197729709542687" target="_blank" rel="nofollow noopener">tiktok.com/@beardedteache
English
0
0
0
30
Sandeep Gangarapu
Sandeep Gangarapu@sandeep_g_kumar·
Some things. I have only one default list in reminders and everything is dumped there. Only exception is Grocery
English
1
0
0
25
Sandeep Gangarapu
Sandeep Gangarapu@sandeep_g_kumar·
TODO stack that works for me. Apple calendar and Apple reminders. I divide tasks into three. 1. Must do on a time and date 2. Good to do on a date 3. Others
English
1
1
0
43
Sandeep Gangarapu
Sandeep Gangarapu@sandeep_g_kumar·
@tomfgoodwin Pachinko on Apple TV. Deep dialogues that made me pause the show and appreciate!
English
0
5
16
861
Tom Goodwin
Tom Goodwin@tomfgoodwin·
I loved homeland , succession and breaking bad I’m now frequently exhausted ( 7 month old baby) and want an amazing tv show that captivates me like these did. And fairly early on in episode 1 Suggestions ?
English
235
0
63
46.4K
(Formerly) Brooklyn Mom
(Formerly) Brooklyn Mom@graciefacelove·
I asked a woman next to me at a coworking space if she could please put headphones on for a meeting and she told me no and that I should get a private office - so now I’m blasting Coco Melon and randomly repeating back full sentences of their meeting.
English
215
1.4K
74.4K
1.9M
Sandeep Gangarapu
Sandeep Gangarapu@sandeep_g_kumar·
@JPEGuin These niche features are awesome but hope the snappiness of the app is not compromised. its still crazy how fast the app feels using.
English
0
0
1
130
Sandeep Gangarapu
Sandeep Gangarapu@sandeep_g_kumar·
a horse cant run in a straight line. a human cant run fast. but a human on a horse can run fast in a straight line
English
0
0
0
19
Sandeep Gangarapu
Sandeep Gangarapu@sandeep_g_kumar·
@JPEGuin filter and sort in explore view. I would love to filter just for example Available on Netflix and sort by tomato meter. Filter by language, year and so on.
English
1
0
0
43
eric zakariasson
eric zakariasson@ericzakariasson·
models im currently using for each mode in cursor: - plan: gpt-5.2 extra high - agent: opus 4.5 & composer-1* - debug: gpt-5.2 extra high - ask: composer-1 *first prompt in opus 4.5, then composer-1
eric zakariasson tweet media
English
119
100
2K
172.1K
Sandeep Gangarapu
Sandeep Gangarapu@sandeep_g_kumar·
One of the first times seeing AI agent handle complex task. Pine AI negotiated xfinity on my behalf and got a non-insignificant discount of $40. There are still hiccups but super impressed to see AI come close to handle a complex task. @PineAiCallAgent 19pine.ai/?utm_source=x&…
English
1
0
0
84
Lenny Rachitsky
Lenny Rachitsky@lennysan·
What's a tool you (or your friend/colleague) vibe coded, that you actually use regularly in your work or life? Looking for examples of real use-cases that stick, vs. just sound good.
English
525
57
1.2K
413.6K
Dmitrii Kovanikov
Dmitrii Kovanikov@ChShersh·
Quant interview question: You press a button that gives your randomly uniformly distributed number between $0 and $100K Each time you press, you have two choices: 1. Stop and take this amount of money 2. Try again You can try 10 times total. When do you stop?
English
807
180
6.8K
1.6M
Sandeep Gangarapu
Sandeep Gangarapu@sandeep_g_kumar·
@GertsDev @theo i have been using search urls to get around this problem. I use raycast to launch these
Sandeep Gangarapu tweet mediaSandeep Gangarapu tweet media
English
0
0
2
234
gertsio ❗
gertsio ❗@gertsio·
Hi @theo, I’m a happy user of t3chat—thanks for creating such an amazing piece of software! 🙌 I might be using it in my own unique way, but I often have multiple tabs open (sometimes 6 or more), and I need each tab to have a specific AI model. Different models are better suited for different tasks, as you know. Right now, the model is synchronized across tabs until the first message is sent. However, I was wondering if it would be possible to switch to a URL-based approach, where the URL updates as soon as the model is chosen. For example, something like t3.chat/chat?model=gem… to store the model selection. This way, the model would be tied to the URL, making it easier to share or bookmark specific configurations. If this approach has technical challenges or trade-offs, I’d love to hear your thoughts—it’d be awesome to understand the reasoning behind the current implementation! Thanks again for all your hard work! 🚀
English
6
2
384
42.4K
Theo - t3.gg
Theo - t3.gg@theo·
What are your biggest frustrations with T3 Chat right now?
English
505
3
800
193.7K