I don't know about that. VCs literally interrogate you like a criminal before they invest. So either all the VCs i came across were a bunch of rude, insensitive pricks, or you're mistaken. I'm genuinely not sure which!

So, i’ve basically started treating X, linkedin and other social platforms as an experiment for my industrial B2B use case. Trying to figure out if I can actually get industrial leads through these platforms(specifically X for now) instead of just getting followers and likes. If you’ve managed to make B2B work through X, linkedin, or something less obvious, I’m curious what you did.

Sending the proposal deck today. Making slide decks is a challenge in itself. You never really know how the other side will receive it, and all you can do is hope the deck communicates the idea well enough.

Call me whatever you want, but the moment I hear “agent this, agent that” in product release videos, I just scroll by. I do it reflexively at this point. It has the same vibe as, “I made a revolutionary new note-taking app.”

Grass always looks greener on the other side. When I was working on B2C, B2B seemed like the sol to all my problems. Now that I’m in B2B, getting that one meeting feels like the end of the world. I guess at the end of the day, it’s your perseverance that makes the difference.

Anybody can spare a muse invite please? Been seeing everybidy talk about it and I'm curious!

I’m 30, based in Bengaluru Founder + deep learning engineer at purnabodhalabs.com Recently incubated at IIT Ropar. Building AI models for real-world industrial & physical systems across climate, pollution, predictive maintenance, and beyond. Always down to meet fellow builders 🚀

Slowing down AI progress makes 0 sense to me. Yes, AI going rogue is a real danger. But how does slowing progress solve that? You’re just postponing the problem by 10 years. What happens after that? And realistically, “slowing down” wont stop the research. It means the public gets less access while the research continues behind closed doors. So all you're really doing is concentrating the power in the hands of a handful of gatekeepers. What problem does that actually solve?

Y’all, things just got VERY real on the B2B side. Had an insane meeting yesterday with an industry(easily worth thousands of crores) and they’re insanely interested in one of our models. We’ve already received an initial quote of interest, the 2nd meeting is already scheduled(from their side ie), and they’re even enquiring about the hardware stuffs. People diss on “model training” until the models start getting industry interest. B2B moves slow, until it doesn’t.

Closed source companies definitely have their moments, but I’m insanely bullish on open source models. No matter how impressive openai, anthropic, and the other closed labs are, open source is catching up in mere days to weeks. Love this competition. And credit where it’s due- china is majority reason why all the closed source labs are facing the heat!

AI wrappers have entered the hardware niche. Put a Raspberry Pi in a toy, connect it to an LLM, slap on a nice enclosure and suddenly it’s disruptive “deep tech.” Using existing tech is fine. Pretending the wrapper is the innovation is where it gets weird.

Something as simple as running YOLO for object detection, recording the screen, and posting it as a “project” gets you 900+ likes and 200k views on linkedin. Meanwhile, I’m out here sharing my adventures with custom models trained from scratch for cell life trajectories, crazy vision models, climate models predicting extreme events, etc., and getting 3 views(two of which are probably my own) And the worst part is that X isn’t any different either! Sigh.

The deep tech startup I’ve been building in stealth for the past 2 months (the actual reason for my absence from social media), purnabodhalabs.com , just got incubated at IIT Ropar ya’ll! We’re building ML systems that understand and model real-world physical systems (climate, environment, predictive maintenance and more), while also working with hardware systems like UAVs. I know this isn’t some crazy huge milestone or anything, but for someone who never in his wildest dreams thought he’d actually be running a company, this is a HUGE deal to me. Two months ago, this wasn’t even an idea, and fast forward to now, we’ve got a formal company structure, we’re incubated at IIT Ropar, and we’re already doing business with actual industries!!!! This whole thing went from literally nothing to something very real ridiculously fast. Feels surreal tbh. And we’re just getting started!

These 15 days away from X made me realize that almost all the buzz around new AI models and other such tech releases lives here on X only. Outside of it, most people don't care (or even know these things happened).

was away for a while! what did I miss?

A century from now, there'll be terabytes of data from people who no longer exist. Another century after that, even more. I wonder how the future will deal with all those digital ghosts. Atleast they won't have to worry about data for training models i guess!

Took 4 days off and now back! Have decided to go all in on building and training models full-time! Predictive maintenance is getting real traction. Also working on other similar industrial use cases, TTS models, and growing YapapaForms. Lots to build!

Vibe coding is becoming the new resume spam. A role gets posted 1,000+ applications arrive 900+ have portfolios full of vibe coded projects Around 50 can actually explain how their projects work ~5 actually understands the architecture, code. Hiring engineers is rapidly becoming a signal extraction problem. Log #007 [Thoughts]

log #006 [Build] Day 24 of building yapapaforms and today built myself a clean and aesthetic dashboard to manage yapapaforms. Having a nice aesthetic dashboard makes everything centralized and easy to manage! Now I understand why people are so particular about aesthetic dashboards!

The marketer's dilemma is real. Every week there's a new "best" growth channel. LinkedIn Reddit SEO X Newsletters So you end up trying all of them, being everywhere, with results nowhere. I'd rather go all in on one platform and fail than spread myself across ten different platforms and never give any of them enough time to work. What's your strategy here? log #005 [Thoughts]