Mohammad Palla

Had fun speaking about 'How to Build a RAG App' at @Microsoft Hyderabad during the #DevRelSquad Gen AI meetup! 🚀 Thanks to everyone who joined and made it an amazing event. #GenAI

I read the answer first, then the question lol. goated feed

@pratykumar @SarvamAI Hoping to meet @archiexzzz in person :)

Spending the day at Sarvam Epoch. Builder Edition.

"We built AI to replace humans." Five minutes later: "Can another AI please deal with this AI?"

@piyush100x @ycombinator @peakxvpartners Congratulations on the launch, i would love to try this

We raised a $15M seed from @ycombinator , @peakxvpartners and some of the killer angels in tech to build vorflux, the absolute best cloud coding agent on earth. (ps: we have already built it. we live on it now.) here's how my 4 months of running my ENTIRE workflow on Vorflux looks like: - haven't run `git clone` locally in 4 months .. not once at all. - claude code even on bypass mode, feels slow af now. - am finally used to running 10 parallel sessions a day .. my mind finally adapted to being fully hands-off and managing agents, developing harnesses, nothing else. - 5,000 commits on the vorflux repo. 99.98% by vorflux itself. vorflux has literally built itself. - most of my time now goes to talking with customers, not writing or managing code. and the absolute best part : my peak monthly usage would look like $30 - 40k in tokens which i get it for absolutely freee by bringing my codex subs

@Rahul_J_Mathur @thealienkind @shahidadnan20 @yaserpatel_x

Alienkind just raised a ₹30 crore round - this company was widely dismissed by observers as “Indiranagar PMF” since its 1st store went viral on Instagram in March 2025. This round was was at a ₹250 crore post money valuation (almost x3 the first round which was at ₹80 crore) The entire funding is contributed by internal investors which is primarily 3 individuals - Ravi Iyer (SVP @ Flipkart), Prakash Sikaria (CEO @ Super Money) and Arpan Sheth (ex Sr Partner @ Bain India) Alienkind’s growth is driven by 3 key factors:

@TheVixhal Every lie is just a cache. The truth eventually invalidates it.

I found out my girlfriend was lying about one guy. For months, she kept saying he was just a friend, and honestly I believed it enough to not make it a daily issue. Not fully, but enough. You know that weird place where something feels off, but you also don't want to become that insecure person who questions every name? Then one night, her phone was on the table and his notification came. “Reached home?” That message alone was not proof of anything. People can ask that. Friends can ask that. I was still trying to be normal about it. But then I remembered she told me she was out with her cousins. So I asked casually, “Who dropped you?” She said, “My cousin.” I asked which cousin. She paused. That pause did more damage than the message. After that, the story started changing as she spoke. First it was cousins. Then it was cousins and friends. Then he was there too, but only for a while. Then it became, “Actually, he dropped me because it was late.” Then somehow I became insecure for even asking. Bro, I was literally watching the probability update in real time. The notification changed it a little. Then the pause changed it even more. By the time the story changed for the third time, the calculation was basically done. That is Bayes theorem. Bayes theorem is what happens when you start with a belief, observe new evidence, and update that belief. Formula: P(H | E) = (P(E | H) × P(H)) / P(E) Where: - H: Hypothesis - E: Event (what you observe) - P(H): Prior probability - P(E): Overall probability of event - P(E|H): Prob. of event if hypothesis was true - P(H|E): Probability hypothesis is true given event (updated belief) Let's take an example and solve it step by step: Suppose your girlfriend is honest 95% of the time and lies 5% of the time. She avoids eye contact 30% of the time when she is honest and 80% of the time when she is lying. If she avoids eye contact, find the probability that she is lying. Step 1: Define Hypotheses - H1: She is honest - H2: She is lying Step 2: Define Event - Event (E) = She avoids eye contact Step 3: Write Priors - P(H1) = 0.95 (honest) - P(H2) = 0.05 (lying) Step 4: Write the Likelihoods Avoids eye contact when honest - P(E | H1) = 0.30 Avoids eye contact when lying - P(E | H2) = 0.80 Step 5: Calculate the Total Probability P(E) = P(E|H1)×P(H1) + P(E|H2)×P(H2) P(E) = (0.30)(0.95) + (0.80)(0.05) P(E) = 0.285 + 0.04 P(E) = 0.325 Step 6: Apply Bayes Theorem - P(H2 | E) = (P(E | H2) × P(H2)) / P(E) - P(H2 | E) = ((0.80)(0.05)) / 0.325 - P(H2 | E) = 0.123 Final Answer: If she avoids eye contact, the probability that she is lying is 12.3%. Congratulations, you have just learned Bayes theorem!

@safishamsii @graphify I can help with Bangalore

I’m looking for a community ambassador at my yc company @graphify. This is mainly about organizing hackathons and doing some devrel to help grow our developer community. Ideally someone with a solid following on socials and good connections around Bangalore or Hyderabad. If that lines up with what you’re into, just dm me. Keen to chat with the right person.

@safishamsii @graphify @TPateeq for Hyderabad

@notsafii 🤣

@PallaMohammad 'auth v2 implementation phase 1' Looks like I have failed 😔

Mondays are for destruction right ?

@notsafii Commit message should be (minor changes)

@PallaMohammad Did something similar but on Friday x.com/i/status/20680…

@championswimmer @sidin The real taste comes from chips, which have the icing of he pastry and the pastry, which has a bit of chowmein in it

@sidin Pineapple pastry, chips and chowmein with 0 veggies or meat in it. Absolute S tier food.

You know what is the most irresistible food in the world? The food at children’s birthday parties. My brain is constantly like: “Can I eat some? Will people judge me? Should I go for it?”

@TPateeq @ycombinator winter arc FTW!!!!

Got an itch while studying in the summer heat Does the season you go through at @ycombinator actually matter to how it turns out? So, ADHD riddled genz that I am, I pulled all ~6k YC companies to see if batch season favors success. yc-seasons.tpateeq.com Winter does win consistently, though barely ( 2% more ) looked only at batches old enough to have matured (6+ yrs), and it's the same every time: fewer shut down, more hit growth, and even more still standing. @paulg @garrytan What do you think? Am I cooking or am I cooked ( legit, it's hot )

Senior engineers aren't the ones who never break production. They're the ones who break it once, learn from it, and build systems so nobody breaks it the same way again.

@arpit_bhayani This becomes even more important in the AI era. Implementation is getting cheaper. Judgment is getting more expensive.

I leave gaps in a design doc on purpose because, if I do not, I am treating my teammates as hands, not brains. The common belief is that a design doc should be detailed enough for an SDE 1 to pick up and implement without thinking. I do not agree with that. If every decision has already been made, the engineer implementing it is just transcribing your thoughts into code, which has now become a commodity. When I leave gaps, two things happen. The engineer feels trusted to make real decisions, not just execute someone else's. And often, they find issues in my thinking that I missed entirely because they are now co-owning it. I am explicit about this when I hand off a doc. I tell them directly that there are gaps and that I expect them to resolve some of them during implementation. This is a very conscious choice I make. See if you would like to try the same. Hope this helps.

@MaxBrodeurUrbas The org chart now compiles on every commit.

@CamiloBAcosta YC founders are now younger than the startups they're disrupting.

The oldest founder in the current YC batch is 31 👀

Everyone's talking about the acquisition. The real story is that coding became valuable enough for a $60B outcome. Software is still eating the world. AI just gave it bigger teeth. 😏

@oneill_c @cursor_ai The naming convention is the real benchmark. Models may improve, but I need to know whether Kobe-7 outperforms Gauss-3.

Was a lot of fun telling @cursor_ai how we manage 128+ agents and making some predictions on the future of language models. Plus hear about how Harry names his agents after mathematicians and I name mine after NBA players, which probably unnecessarily hobbles mine in comparison but I love it too much to stop