📣Thrilled to announce I’ll join Carnegie Mellon University (CMU Engineering & Public Policy & Language Technologies Institute | @CarnegieMellon) as an Assistant Professor starting Fall 2026! Until then, I’ll be a Research Scientist at AI at Meta FAIR in SF, working with Kamalika Chaudhuri’s amazing team on privacy, security, and reasoning in LLMs!

When an old man dies, a library burns to the ground

How to train a model that actually understands both audio and text like Voxtral from Mistral AI? Here is a quick video walkthrough of the paper.

Swalpa Math problem idu. Try this instead

Introducing our new work: 🚀Mixture-of-Recursions! 🪄We propose a novel framework that dynamically allocates recursion depth per token. 🪄MoR is an efficient architecture with fewer params, reduced KV cache memory, and 2× greater throughput— maintaining comparable performance!

Great new blog post by Sebastian Raschka about the evolution and advancement of different LLM architectures!

Excited to start as a Research Scientist Intern at Meta, in the GenAI Red Team, where I will keep working on AI agents security. I'll be based in the Bay Area, so reach out if you're around and wanna chat about AI security!

Hurwitz's theorem is surprising

MathMatize Memes Here's a theorem that should blow your mind: en.wikipedia.org/wiki/Kolmogoro…

Advances in Computer Vision from MIT Open Courseware is available freely and is one of the fantastic courses.

An introduction to information geometry in 10 pages: "The Many Faces of Information Geometry" Open access PDF

Why do FFNs use ReLU instead of more precise ones like Exp? "We propose the following hypothesis: A kernel with lower retrieval precision encourages a more polysemantic key–value memory: multiple unrelated facts can be stored under the same key space" Great and inspiring read!

Wild paper They prove (!!) a transformer block (Attn + MLP) running on prompt Outputs the same logits with no prompt If MLP weights updated by vector: W′ = W + ΔW Calc from attn latent: ΔW = (W·Δa) × (A(x)ᵀ / ‖A(x)‖²) Given prompt: Δa = A(C, x) − A(x) Fucking fine tuning.

In-context learning is just gradient descent without explicit training! This paper "Learning without training: The implicit dynamics of in-context learning" shows that ICL can be mathematically interpreted as an implicit low-rank weight update during inference.

Fuck ML tutorials. This is a collection of 300 ML system design case studies in real world, from Stripe, Spotify, Netflix, Meta, etc. Perfect for interviews and to learn how it’s done in the battlefield. Wish there was a similar thing for agents!

LLMs can now self-optimize. A new method allows an AI to rewrite its own prompts to achieve up to 35x greater efficiency, outperforming both Reinforcement Learning and Fine-Tuning for complex reasoning. UC Berkeley, Stanford, and Databricks introduce a new method called GEPA

India is nowhere in the NLP map. 😢 We, cumulatively, need to push ourselves hard. ACL 2025 #nlproc

Geometric-Mean Policy Optimization Using geometric mean for the importance ratio, similar to GSPO (arxiv.org/abs/2507.18071).

Build a Large Language Model from scratch! This repository contains the code for developing, pretraining, and finetuning a GPT-like large language model. 100% Free & Open Source

An amazing video, our national animal and bird, together in one frame! A perfect symbol of India's vibrant spirit. Wishing everyone a Happy Independence Day. आप सभी को स्वतंत्रता दिवस की हार्दिक बधाई एवं शुभकामनाएं, जय हिंद। 🇮🇳 VC: Rakesh Bhatt #IndependenceDay #JaiHind