🧠 Self-Hosted AI Assistant with Custom Model (Elion)

📖 Overview

This project demonstrates the deployment of a fully self-hosted Large Language Model (LLM) environment using Ollama and Open WebUI on Linux.

The system includes:

• Local LLM deployment
• Custom model creation using Ollama Modelfile
• API configuration and testing
• WebUI integration
• Troubleshooting of port and service conflicts
• Fully offline AI interaction without cloud APIs

The custom model Elion was built on top of a base model and configured with specialized system behavior and generation parameters.

---

🎯 Project Objectives

• Deploy Ollama on Linux
• Run and manage local LLM models
• Create a custom AI model (Elion)
• Integrate Ollama API with Open WebUI
• Debug and resolve API connection issues
• Successfully interact with a self-hosted AI assistant

---

🛠 Technologies Used

• Linux (Ubuntu)
• Ollama
• Open WebUI
• REST API
• Curl
• Git & GitHub

---

⚙️ Installation & Setup

1️⃣ Install Ollama

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

Verify installation:

ollama --version

---

2️⃣ Start Ollama Server

ollama serve

Server runs on:

http://127.0.0.1:11434

---

3️⃣ Pull Base Model

ollama pull qwen2.5:3b-instruct-q4_K_M

---

🧠 Custom Model Creation – Elion

📌 Step 1: Create Modelfile

nano Modelfile

Example Modelfile:

FROM qwen2.5:3b-instruct-q4_K_M

SYSTEM You are Elion, a professional AI assistant specialized in cybersecurity, Linux troubleshooting, and system diagnostics.

PARAMETER temperature 0.7
PARAMETER num_ctx 4096
PARAMETER top_p 0.9

---

📌 Step 2: Build Custom Model

ollama create elion -f Modelfile

---

📌 Step 3: Verify Model Creation

ollama list

Expected output:

elion:latest
qwen2.5:3b-instruct-q4_K_M

---

📌 Step 4: Test via API

curl http://127.0.0.1:11434/api/generate -d '{
  "model": "elion",
  "prompt": "Explain Linux port conflicts",
  "stream": false
}'

---

🌐 Open WebUI Integration

1. Launch Open WebUI
2. Navigate to:
   Admin Panel → Settings → Connections
3. Add Ollama connection:

http://127.0.0.1:11434

4. Save and refresh
5. Select elion:latest from model dropdown

---

🔍 Troubleshooting

Issue: Port 11434 already in use

sudo lsof -i :11434
sudo killall ollama

---

Issue: "Failed to fetch models"

• Use http://127.0.0.1:11434 instead of localhost
• Ensure ollama serve is running
• Disable "Cache Base Model List" in WebUI settings

---

Issue: Server connection error

• Confirm Ollama is listening:

curl http://127.0.0.1:11434/api/version

---

📷 Screenshots

Add these images inside /images folder:

![Ollama Server Running](images/ollama-server.png)
![Elion Modelfile](images/modelfile.png)
![Model Creation Process](images/elion-creation.png)
![WebUI Connection Setup](images/webui-connection.png)
![Successful AI Chat](images/chat-response.png)

Recommended screenshots:

• Terminal showing ollama serve
• Modelfile content
• ollama create elion
• WebUI connection configuration
• Chat interaction with Elion

---

🚀 Results

Successfully deployed and configured a self-hosted AI assistant with:

• Custom model behavior
• Local API communication
• Web interface integration
• Fully offline operation
• Resolved real-world service and networking issues