Back to Portfolio

Free LLMs - Local AI Interface

AI & Machine Learning View on GitHub

A modern, production-ready web application that provides a beautiful interface for interacting with local Large Language Models through Ollama.

Problem & Solution

The Problem

Cloud-based AI services raise concerns about data privacy, require internet connectivity, and often come with API costs. While Ollama provides local LLM capabilities, it lacks a user-friendly interface for non-technical users and doesn't offer advanced features like chat management, model comparison, or hyperparameter tuning.

The Solution

Free LLMs bridges this gap with a WhatsApp-inspired interface that makes local AI accessible to everyone. It provides complete control over your data with zero cloud dependencies, supports multiple models with customizable hyperparameters, and includes advanced features like DeepSeek reasoning visualization, real-time streaming responses, and comprehensive chat management—all running entirely on your local machine.

Technologies Used

Frontend

Next.js 14 TypeScript Tailwind CSS React

AI & Backend

Ollama Server-Sent Events Node.js

Key Achievements

Intuitive WhatsApp-Inspired Interface

Designed a familiar 1/3-2/3 split layout with chat list and conversation view, complete with unread indicators, real-time updates, and a modern dark theme that makes AI interaction feel natural and accessible.

DeepSeek Reasoning Visualization

Implemented real-time visualization of AI reasoning processes in collapsible sections, allowing users to see how models think through problems—perfect for learning, debugging, and building trust in AI decisions.

Advanced Model Management

Built an intelligent model downloader with variant selection, real-time progress tracking, and hardware requirement indicators. Users can browse, download, and manage models from Ollama's library with a single click.

Complete Privacy & Local Storage

Architected a zero-cloud-dependency system with all conversations stored locally in JSON format. Includes chat persistence, export to Markdown, and search functionality—ensuring complete data ownership and privacy.

Fine-Grained Control

Developed per-chat hyperparameter customization (temperature, top-p, top-k, max tokens) and system message configuration, giving users precise control over AI behavior for different use cases.

Screenshots

Free LLMs Interface

WhatsApp-Inspired Chat Interface

DeepSeek Reasoning

DeepSeek Reasoning Visualization

Model Management

Advanced Model Management