Krish Dasgupta

Forever Learner | Building Reinforcement Learning Systems | Healthcare | Robots and Brains | Graph ML for Health | Privacy and Decentralized AI

[Krish Dasgupta @officialKrishD]

It was fun presenting at NODES 2025 ! Thanks @neo4j
@JulesHitz

[Krish Dasgupta @officialKrishD]

We believed in the intersection of Graphs and Language models , long before it was cool and graph RAG existed.
We created our own graph embeddings and used them to relate to better decision making process.
Join us on 6th Nov @neo4j and witness our work !
https://t.co/cw7Yzb8bzJ

[Krish Dasgupta @officialKrishD]

@gayathrithedev Eradication of human longingness.

[Krish Dasgupta @officialKrishD]

@tereza_tizkova @anthonyronning Congratulations @tereza_tizkova !!!!

[Krish Dasgupta @officialKrishD]

@ramin_m_h Secretly he uses liquid AI at all his small edge devices 🤪

[Krish Dasgupta @officialKrishD]

@liquidai Beautiful concept !!!!! Can’t wait to try - been so far only working with Presidio !

[Krish Dasgupta @officialKrishD]

@kunalvg Hoping so.

[Krish Dasgupta @officialKrishD]

@0xSero Build and use these kind of projects -
https://t.co/3rmTNVgoEJ

Also quite happy that you are focusing on private and local AI.

[Krish Dasgupta @officialKrishD]

“Don’t be a Homelander.” This is how your agents feel and act when your tokens are exhausted.
So, go local. Build upon the compute. Make the most out of local models.
In courtesy of @TheBoysTV.
Issued in the interest of private compute and local models @huggingface

[Krish Dasgupta @officialKrishD]

6 GB VRAM - 16 Cores - 16 GB RAM
one machine that routes models and also gives GPU precendence and CPU precedence - MedGemma1.54B Q8, Gemma4E4B, Gemma4E2B, Presidio Analyzer and Anonymizer, Gemma3 270M-it_q8 finetuned, and Qwen3.5_0.8B-Q8.
All these models takes precendence on tasks.
The Video Processing is handled by Gemma4E4B on a separate 16GB VRAM machine when needed. 0nly for video processing. Everything else runs via the 6GB VRAM machine.

TLDR : It is upto you how you juice out your compute. It is just there, Max it out, on inference.

[Krish Dasgupta @officialKrishD]

@hthieblot If you know what ASL means. That was a time. One golden time.

[Krish Dasgupta @officialKrishD]

@sophiamyang Huge Congratulations!!!!! Sophia ! This is awesome !

[Krish Dasgupta @officialKrishD]

@MicrosoftAI @jm_alexia Open source ? Weights ? HF LINK !!!!!!

[Krish Dasgupta @officialKrishD]

@sundarpichai Start a hackathon :-)

[Krish Dasgupta @officialKrishD]

@GoogleIndia Honestly - just honestly tell me this - what was the count of query seeking Windows OS serial key 2003-2009

[Krish Dasgupta @officialKrishD]

@ariG23498 I have been using a coupled Gitlab self hosted in my local servers setup.

[Krish Dasgupta @officialKrishD]

@kaggle Hackathon soon ? Eh @Kaggle ?

[Krish Dasgupta @officialKrishD]

The scavengers of LinkedIn are at it again.
If you are in cybersecurity domain, you might have heard about the potential and disastrous leakage of answer sheets by one of the most critical examination bodies of India. [ CBSE ]
LinkedIn Scavengers - Talking about the leak without giving credit to the actual folks who found it. That’s why we have limited genuine raw talent. Since many are just trying to copy / paste known big facts without giving the due credit.
Credit Hogging is a skill that should be added on LinkedIn.

[Krish Dasgupta @officialKrishD]

@sampritibh The brilliant Sampriti.

[Krish Dasgupta @officialKrishD]

If your agent can do 1/2 of the jobs that it is doing now, but without internet and low resource compute - you built yourself a moat.
Else it will be still a glorified API call only this time instead of a wrapper - they call your agent a tad an empty suit.
TLDR : works on local models with low compute ? You built it right.

[Krish Dasgupta @officialKrishD]

AI OS.

Natively supports all inference types, memory swapping with base model - that inherently helps with all tools and action.
Completely local running STS pipeline.
Built with base 1B model for instruction following and tooling.
Natively supports robotics .
Fork of Linux .

Soon.