I Built a Private, Local-First AI Assistant with Flask

Ramesses_2 — Fri, 24 Apr 2026 11:23:07 +0000

The Goal: I wanted an AI assistant that doesn't save my logs to the cloud and utilize my hardware.

Though it's quite slow as of now, I believe with further optimization and advancement in local silicon, edge AI would inevitably become the next big thing in the very near future.

The Tech: I used Flask, TinyLlama, and Python-dotenv for security. I also took the help of Claude and Copilot wherever the code was repetitive and well beyond my current grasp on Python.
I am a college freshman (about to become a sophomore), but I seek merciless feedbacks. Do provide me honest feedbacks about this project.

Github Repository

sounak1410 / Web-Based-Edge-AI-

A personalized Small Language Model that runs locally on your device's hardware without having to access the internet or the cloud

Y Assistant

A privacy-focussed local AI chatbot powered by TinyLlama

Y is a lightweight web-based AI assistant designed to run entirely on your local machine. It features a custom architecture demo, a secure login system, and per-session memory, ensuring that your conversations stay private and contextual. Created by S (My pseudonymn).

Features

No data leaves your machine. It runs on TinyLlama 1.1B.
Contextual Memory: Remembers the last 10 messages for a natural conversation flow.
Secure Access: Protected by a customizable password system.
Privacy-First: No permanent logs are stored; session data is cleared on request.
Architecture Demo: Includes a raw GPT-2 initialization script to show how LLMs are structured.

Installation

Install my-project with npm

  # Clone the repository
git clone https://github.com/sounak1410/Web-Based-Edge-AI-.git

# Enter the directory
cd Web-Based-Edge-AI-

# Install dependencies
pip install -r requirements.txt

# Run the application
python Edge.py

Chatting with Y

Open your browser to http://127.0.0.1:5000.
Enter the…

View on GitHub