Manav Gupta (@tensor___boy) header
Manav Gupta (@tensor___boy) profile picture

i build ai and show you exactly how it works. cto @ Cevi AI (US Healthcare). 2x TEDx. 245k+ on IG. collabs: manav@tensorboy.com

Singapore
Joined December 2021
Photo by Manav Gupta (@tensor___boy)Photo by Manav Gupta (@tensor___boy)Photo by Manav Gupta (@tensor___boy)Photo by Manav Gupta (@tensor___boy)Photo by Manav Gupta (@tensor___boy)Photo by Manav Gupta (@tensor___boy)Photo by Manav Gupta (@tensor___boy)Photo by Manav Gupta (@tensor___boy)Photo by Manav Gupta (@tensor___boy)Photo by Manav Gupta (@tensor___boy)
i started Tensor Boy on my roommate's phone. friends held a bulb for light. 50 views a post. then post #49 got me 10K followers overnight. now: 245K+ on IG 2x TEDx cto @ Cevi AI, voice agents for US clinics Collabs of Apple, Google, Notion, Warp, and YCs if i can do it from Etawah, you can do it from anywhere. here i show you exactly how i build in AI. follow along 🫂
2
11
440
tell me a better way to end a thursday than playing football with the @JoinEdgeCity boys
6
16
520
Morning gaang Trying paper clip … Its kinda overwhelming So removed 80% of the buttons and shit. I don’t know … my adhd don’t like more than 3 things at a time
1
123
Its finally here.. The real system..that actually works...when it comes to agents and actions.
144
Yooo @bot
103
I might need to spend $1999 for research and development purposes.
99
This is how… peak male life looks like Waiting for astra and claude to get the shit done
1
99
Notion is the classic example of why unwanted complexity brings paralysis. All in all don’t give crazzy choices to the dumb crowd…they will die in between
3
668
nvidia is reportedly backstopping openai to the tune of ~$250b inside a 10-gigawatt ohio data center with softbank. the chipmaker is now basically financing its own biggest customer. at what point is this just one giant company wearing three trench coats
122
the CXMT pop is bigger than one hot ipo. china's whole push to make its own memory chips just got a 466% market blessing on day one. felt like watching the chip war quietly get a very loud scoreboard, honestly
124
apparently the new move in ai is "thrift-maxxing": companies quietly mixing cheap chinese models with openai/anthropic just to survive the bill. we went from "use the best model" to "use the cheapest one that doesn't embarrass us" real fast. the big labs cannot be loving this trend lol
1
103
repo of the day: litellm. one api to call 100+ models, openai, anthropic, gemini, the cheap chinese ones, all through the same interface, with cost tracking built in. feels perfectly timed for everyone trying to route around their inference bill right now. github.com/BerriAI/litellm
1
124
a chinese memory-chip maker called CXMT just debuted in shanghai and popped ~466% in a single day. it's now the most valuable company listed in china. a chip company most people outside the industry had never heard of. china basically said "we'll build our own memory" and the market went feral for it
1
126
a chinese memory-chip maker called CXMT just debuted in shanghai, popped ~466% in a day, and became the most valuable company listed in china. here's why it matters beyond the stock
1
116
and in the middle of that, china's homegrown memory champion goes public and the market immediately values it above everything else listed there
1
13
export controls were meant to slow china down. a 466% debut looks a lot more like they sped up the "build it ourselves" plan instead. wild to watch in real time
12