--- title: "Gautam Gaurisaria (@gautamgaurisaria) | Hashnode" description: "Learning....always. Read the latest articles by Gautam Gaurisaria on Hashnode." url: "https://hashnode.com/@gautamgaurisaria" --- @gautamgaurisaria # Gautam Gaurisaria @gautamgaurisaria · India · Joined May 2024 Learning....always ### [The Authenticity Paradox: Why AI Can Write a Joke but Can’t Be Samay Raina](https://devopsdayonegautam.hashnode.dev/the-authenticity-paradox-why-ai-can-write-a-joke-but-can-t-be-samay-raina?utm_source=hashnode&utm_medium=feed) I spent the last few weeks deep in the weeds of AI limitations. Jailbreaks, inference quirks, why models hallucinate when pushed too far. Core DL stuff. And somewhere between trying to break LLMs and 3 min read · [devopsdayonegautam.hashnode.dev](https://devopsdayonegautam.hashnode.dev/the-authenticity-paradox-why-ai-can-write-a-joke-but-can-t-be-samay-raina?utm_source=hashnode&utm_medium=feed) [Mar 22](https://hashnode.com/posts/the-authenticity-paradox-why-ai-can-write-a-joke-but-can-t-be-samay-raina/69bfbfb74a1e513e41bda1be) ### [This Project Took 20 Minutes. That’s Not the Point.](https://devopsdayonegautam.hashnode.dev/this-project-took-20-minutes-thats-not-the-point?utm_source=hashnode&utm_medium=feed) I recently built a small web project using GitHub Copilot AI Agents. It took about 15–20 minutes, maybe 2–3 prompts, and most of the visible structure was done. Let me get this out of the way early:this is not an achievement. What is worth writing ab... 3 min read · [devopsdayonegautam.hashnode.dev](https://devopsdayonegautam.hashnode.dev/this-project-took-20-minutes-thats-not-the-point?utm_source=hashnode&utm_medium=feed) [Jan 25](https://hashnode.com/posts/this-project-took-20-minutes-thats-not-the-point/6976266f22212340ab5feffe) ### [Why Meta’s plays (Scale + Manus) matter: habit, data, and agents — not just better weights](https://devopsdayonegautam.hashnode.dev/why-metas-plays-scale-manus-matter-habit-data-and-agents-not-just-better-weights?utm_source=hashnode&utm_medium=feed) Meta’s AI strategy in 2025 is no accident. It follows a two-pronged playbook: own the data & pipeline that make models reliably useful, and own the agentic interfaces that turn model outputs into repeatable user habits. The Scale AI move sealed a ver... 6 min read · [devopsdayonegautam.hashnode.dev](https://devopsdayonegautam.hashnode.dev/why-metas-plays-scale-manus-matter-habit-data-and-agents-not-just-better-weights?utm_source=hashnode&utm_medium=feed) [Jan 1](https://hashnode.com/posts/why-metas-plays-scale-manus-matter-habit-data-and-agents-not-just-better-weights/69560f65d7013bf812a5dda6) ### [Meta Didn’t Buy Scale AI. It Did Something Smarter.](https://devopsdayonegautam.hashnode.dev/meta-didnt-buy-scale-ai-it-did-something-smarter?utm_source=hashnode&utm_medium=feed) When people talk about big tech acquisitions, they usually imagine clean headlines. Company A acquires Company B for X billion dollars. Regulators react. Antitrust lawyers wake up. The story is easy to tell. What Meta did with Scale AI does not fit t... 3 min read · [devopsdayonegautam.hashnode.dev](https://devopsdayonegautam.hashnode.dev/meta-didnt-buy-scale-ai-it-did-something-smarter?utm_source=hashnode&utm_medium=feed) [Dec 26, 2025](https://hashnode.com/posts/meta-didnt-buy-scale-ai-it-did-something-smarter/694e206e93468f05e9635330) ### [I Was Proven Wrong About the AI Hardware Race](https://devopsdayonegautam.hashnode.dev/i-was-proven-wrong-about-the-ai-hardware-race?utm_source=hashnode&utm_medium=feed) A few days ago, I took a clear position. I argued that purpose-built AI architectures like LPUs and wafer-scale systems would eventually outcompete GPUs. From that, I went a step further and predicted that NVIDIA might not be the ultimate winner of t... 3 min read · [devopsdayonegautam.hashnode.dev](https://devopsdayonegautam.hashnode.dev/i-was-proven-wrong-about-the-ai-hardware-race?utm_source=hashnode&utm_medium=feed) [Dec 26, 2025](https://hashnode.com/posts/i-was-proven-wrong-about-the-ai-hardware-race/694e1f9e9bc9439c52e3e6f0)