Build Your Own Offline AI Code Assistant on 4GB RAM VPS: CodeGPT & CodeLlama with Ollama + VS Code

Introduction to Offline AI Code Assistants

In the rapidly evolving landscape of software development, AI-powered code assistance has become an essential tool for developers seeking to improve productivity and code quality. However, many developers hesitate to adopt cloud-based solutions due to privacy concerns, API costs, and dependency on internet connectivity. Fortunately, there is a compelling alternative: building your own offline AI code assistant.

This comprehensive guide will walk you through the process of setting up a private AI coding assistant that runs entirely on your own infrastructure. By leveraging Ollama, an open-source runtime for running large language models, and integrating it with Visual Studio Code, you can create a powerful coding companion that operates completely offline while maintaining full control over your data.

Understanding the Technology Stack

What is Ollama?

Ollama is an open-source platform that enables developers to run large language models locally on their machines or servers. It provides a simple yet powerful interface for deploying and managing AI models, making it accessible even to those without extensive machine learning expertise. Ollama supports various models, including the popular CodeLlama family, which is specifically designed for code generation and understanding.

Why VPS with 4GB RAM?

While running AI models locally on a personal computer is possible, deploying them on a Virtual Private Server (VPS) offers several distinct advantages. A VPS provides consistent performance, 24/7 availability, and the ability to access your AI assistant from any location. A 4GB RAM configuration represents an excellent balance between cost and capability, allowing you to run efficient models without significant financial investment.

Prerequisites and Initial Setup

Before beginning the installation process, ensure you have the following requirements in place:

A VPS with at least 4GB of RAM (recommended: 4GB or more)
Ubuntu 20.04 LTS or later operating system
SSH access to your VPS
Basic command-line knowledge
Visual Studio Code installed on your local machine

Step-by-Step Installation Guide

Step 1: Preparing Your VPS Environment

Begin by connecting to your VPS via SSH and updating the system packages. Execute the following commands to ensure your server is up to date:

sudo apt update && sudo apt upgrade -y

This process may take several minutes depending on your server configuration. Once completed, you will have a stable foundation for installing Ollama and its dependencies.

Step 2: Installing Ollama

Ollama provides a straightforward installation script that handles most of the heavy lifting. Run the following command to install Ollama:

curl -fsSL https://ollama.com/install.sh | sh

The installer will automatically download the necessary components and configure your system. After installation completes, verify that Ollama is properly installed by checking its version:

ollama --version

Step 3: Selecting and Installing Code Models

With Ollama installed, you can now pull the code generation models. CodeLlama is an excellent choice for code assistance, offering various sizes to accommodate different RAM constraints. For a 4GB RAM VPS, we recommend starting with the 7B parameter model, which provides a good balance between capability and resource consumption.

To download and install CodeLlama, execute:

ollama pull codellama:7b

Alternatively, you can explore other models such as Stable Code or DeepSeek-Coder, which are specifically optimized for code generation tasks. Each model has different characteristics, so feel free to experiment to find the best fit for your needs.

Step 4: Configuring Ollama for Remote Access

By default, Ollama binds to localhost only. To access your AI assistant from external clients, you need to configure it to listen on your server's IP address. Set the OLLAMA_HOST environment variable to allow remote connections:

export OLLAMA_HOST=0.0.0.0:11434

For persistent configuration, add this variable to your system's profile. Additionally, ensure your firewall allows traffic on port 11434 if you have one configured.

Integrating with Visual Studio Code

Installing the Ollama VS Code Extension

Visual Studio Code offers excellent extensibility, and several extensions enable integration with Ollama. The most popular option is the Ollama Extension available in the VS Code marketplace. To install it:

Open Visual Studio Code on your local machine
Navigate to the Extensions panel (Ctrl+Shift+X)
Search for "Ollama"
Click Install on the appropriate extension

Configuring the Extension

After installation, configure the extension to connect to your VPS. Open the extension settings and specify your server's IP address and port number. The typical configuration looks like:

Base URL: http://your-vps-ip:11434
Model: codellama:7b

Once configured, you can start using AI-assisted coding directly within your VS Code environment. The extension provides inline suggestions, code completion, and chat functionality powered by your local AI model.

Optimizing Performance on 4GB RAM

Model Selection Strategies

Running AI models on limited RAM requires careful optimization. Here are some strategies to maximize performance:

Choose smaller parameter models: The 7B models typically require 4-8GB of RAM for inference, while 3B models can run comfortably within 2-4GB.
Implement model quantization: Quantized models use less memory with minimal quality loss. Ollama supports various quantization levels.
Limit concurrent requests: Process requests sequentially to avoid memory exhaustion.

System Optimization Tips

Enhance your VPS performance with these additional optimizations:

Enable swap space to provide additional virtual memory
Close unnecessary background processes
Use lightweight server configurations
Monitor resource usage with tools like htop or glances

Security Considerations

When exposing Ollama to network access, security should be a primary concern. Implement the following best practices:

Use authentication: Consider implementing an authentication layer if accessing from public networks
Enable HTTPS: Configure SSL/TLS encryption for secure communications
Firewall rules: Restrict access to trusted IP addresses only
Regular updates: Keep Ollama and your operating system updated with security patches

Cost Analysis and Benefits

Financial Advantages

Building your own offline AI assistant offers significant cost benefits compared to cloud-based alternatives. Consider the following comparison:

Cloud API costs: Traditional AI coding assistants charge per token, which can accumulate rapidly
VPS costs: A 4GB RAM VPS typically costs $10-20 per month
One-time investment: After initial setup, there are no additional per-use charges

Additional Benefits

Beyond cost savings, your self-hosted solution provides:

Complete data privacy: Your code never leaves your infrastructure
Offline capability: Work without internet connectivity
Customization: Fine-tune models to your specific needs
No rate limits: Unlimited usage without restrictions

Troubleshooting Common Issues

Memory-Related Problems

If you encounter out-of-memory errors, try these solutions:

Reduce the model size or switch to a more quantized version
Increase swap space on your VPS
Restart the Ollama service to free up memory
Close other applications consuming RAM

Connection Issues

For connection problems between VS Code and your VPS:

Verify firewall rules allow traffic on port 11434
Confirm the OLLAMA_HOST environment variable is set correctly
Test connectivity using curl or telnet
Check that Ollama service is running (systemctl status ollama)

Conclusion and Future Directions

Building a private AI code assistant on a 4GB RAM VPS represents a significant milestone in taking control of your development environment. By following this guide, you have created a fully functional, offline-capable coding companion that rivals commercial alternatives while offering superior privacy and cost-effectiveness.

As you become more comfortable with your setup, consider exploring advanced topics such as fine-tuning models on your codebase, integrating additional tools, or scaling your infrastructure. The open-source nature of Ollama ensures continuous improvements and new features will be available as the technology evolves.

Start your journey today and experience the freedom of having your own AI coding assistant—one that works exactly when and how you need it, without compromise.