Ayush Ranjan

AI Engineer | RAG | AI Agents & LLMs

[Ayush Ranjan @ayusrjn]

I am building an ai coding agent. "Lambda"
I have realised how important it is to protect your context window.
https://t.co/gCc9zlMgul

[Ayush Ranjan @ayusrjn]

@theishanvyas Check out my GitHub - https://t.co/3Jk3B05Q5i
One of my project - https://t.co/JwWXobepJV
I specialize in multimodal agents - https://t.co/5akXYyLp6a

[Ayush Ranjan @ayusrjn]

@neural_avb I am building my custom AI coding agent like claude code
https://t.co/2KeTu7zqV5

[Ayush Ranjan @ayusrjn]

I am building my own coding agent.
Try it here
https://t.co/2KeTu7zqV5

[Ayush Ranjan @ayusrjn]

new coding agent on market
"lambda"
https://t.co/2KeTu7zqV5

[Ayush Ranjan @ayusrjn]

vibe coding with my own coding agent
"lambda"

[Ayush Ranjan @ayusrjn]

Managing the context window is the real challenge.

[Ayush Ranjan @ayusrjn]

I am building an ai coding agent. "Lambda"
I have realised how important it is to protect your context window.
https://t.co/gCc9zlMgul

[Ayush Ranjan @ayusrjn]

Working on Image Processing toolbox for Scilab @Scilab
IPCV has been long outdated

[Ayush Ranjan @ayusrjn]

@EXM7777 To develop with Gemma 4
With @cactuscompute , you can create app which can use the gemma 4 model for local use cases

[Ayush Ranjan @ayusrjn]

@Ashutosh_7x7 Same bro gave everything this year,
But now all a man can do is hope

[Ayush Ranjan @ayusrjn]

@AroraBhavyam AI

[Ayush Ranjan @ayusrjn]

I ran two vLLM server with gemma 2b on kaggle,
So I don't have to pay any bill

[Ayush Ranjan @ayusrjn]

I have automated my WhatsApp with openclaw to flirt with my crushes

So far I have a meeting with one at 10 tomorrow.

See how ⬇️

[Ayush Ranjan @ayusrjn]

@aditnotfound FORGE

[Ayush Ranjan @ayusrjn]

@Suryanshti777 DL

[Ayush Ranjan @ayusrjn]

Deployed qweb 2.5 3B model with vLLM
dealt with a lot of package compatibility issues due to older gpu and cuda. But figured out eventually.
Gives around 35-40 tokens/sec

[Ayush Ranjan @ayusrjn]

I will be using 7B self hosted model

[Ayush Ranjan @ayusrjn]

Building a timeline-aware, hybrid RAG system over public government audit data.
Focus: retrieval quality, metadata-aware search, reranking, and evaluation.

Goal: Make policy documents actually searchable and analyzable.

[Ayush Ranjan @ayusrjn]

Deepseek moment for india,
Sarvam ai lauches it's foundational model

[Sarvam for Developers @SarvamForDevs]

The future of India Starts Here.

Join Now.