Madhav Kanda @madhav_kanda_

MSCS @UofIllinois | Prev @MSFTResearch , @Azure, @IITGN, @AaltoUniversity, @iiscbangalore, @spyne_ai madhav-kanda.github.io United States Joined July 2022

Tweets

18
Followers

70
Following

236
Likes

64

Madhav Kanda @madhav_kanda_

a week ago

Running a language model locally on my MacBook Air with low latency — MLX + quantization are magic 🚀

0 0 6 84 0

PL/FM/SE at Illinois @plfmse

3 weeks ago

🚀 Introducing Structured LLM, a new framework for making large language models more aligned, useful, and efficient. 👉 Check it out here: structuredllm.com

1 7 18 1K 6

Ashwini Vaishnaw @AshwiniVaishnaw

4 weeks ago

5️⃣ L&T Semicon + IIT Gandhinagar & C-DAC ✔️ Development of Make-in-India Secure Chip, with a Smart Operating System, designed to power the next generation of digital identity solutions. ➡️ Applications: e-passports, national IDs & cryptographic products.

3 120 715 52K 15

Download Video

Madhav Kanda @madhav_kanda_

4 months ago

🚀 Excited to be attending YC’s AI Startup School in SF! Looking forward to learning, building, and connecting with others in the AI space. If you’re attending too, let’s connect!!

0 0 3 206 0

Madhav Kanda @madhav_kanda_

4 months ago

Thrilled to share that I’m joining @MSFTResearch in Redmond as a Research Intern! I’ll be working with both the Azure Research Systems Group (@Azure) and @MSFTResearch. Excited to learn, grow, and contribute! 🚀🧠💻

2 0 13 633 0

Sagnik Mukherjee @saagnikkk

4 months ago

🚨 Paper Alert: “RL Finetunes Small Subnetworks in Large Language Models” From DeepSeek V3 Base to DeepSeek R1 Zero, a whopping 86% of parameters were NOT updated during RL training 😮😮 And this isn’t a one-off. The pattern holds across RL algorithms and models. 🧵A Deep Dive

16 129 884 179K 886

Download Image

Shubham Ugare @shubhamdugare

5 months ago

Meet us at #ICLR2025 next week to hear about IterGen! We introduce grammar-guided navigation for LLM generation --- precise forward and backward control to improve semantic accuracy.