Build Your Own Local AI Rig – Step-by-Step Guide

This guide walks you through building and installing a complete offline-capable AI workstation for under $1500 CAD. Everything runs locally — no cloud access or API keys required.

With this setup, you'll be able to run:

✅ LLaMA 3.2 8B and other models (via Ollama)
✅ DeepSeek and Qwen for coding and chat
✅ Whisper for speech-to-text
✅ Piper for local text-to-speech
✅ Open WebUI for visual chatting with your models
✅ n8n for automating LLM workflows

🧱 Step 1: Gather Your Hardware (Under $1500 CAD)

Component	Suggested Model	Est. Price (CAD)	Link
CPU	AMD Ryzen 7 7700X	~CA$409	Buy
GPU	NVIDIA RTX 4070 (12GB)	~CA$959	Buy
RAM	32GB DDR5 (5600MHz)	~CA$143	Buy
Storage	1TB NVMe Gen4 SSD	~CA$110	Pick brand
Motherboard	B650 ATX	~CA$200	Pick brand
PSU	650W+ Gold-rated	~CA$120	e.g., RM650x
Case + Cooler	Mid-tower + tower cooler	~CA$90	Pick style

Total: ~CA$1900 — you can stay under $1500 with used GPU or budget motherboard.

🛠️ Step 2: Assemble the PC

Install CPU, cooler, and RAM onto motherboard.
Mount motherboard in case.
Install SSD, GPU, and PSU.
Connect all power and data cables.
Power on and enter BIOS — set boot priority.
Install Ubuntu 22.04 LTS from USB.

🔧 Step 3: Update Your System

sudo apt update && sudo apt upgrade -y

🤖 Step 4: Install Ollama (LLM Engine)

curl -fsSL https://ollama.com/install.sh | sh
ollama run llama3
ollama run deepseek-coder:8b
ollama run qwen:7b

🌐 Step 5: Install Open WebUI

git clone https://github.com/open-webui/open-webui.git
cd open-webui
./scripts/docker-run.sh

Open http://localhost:3000 in your browser.

⚙️ Step 6: Install n8n

docker run -it --rm \
  -p 5678:5678 \
  -v ~/.n8n:/home/node/.n8n \
  n8nio/n8n

Open http://localhost:5678

🗣 Step 7: Install Whisper (Speech-to-Text)

git clone https://github.com/ggerganov/whisper.cpp
cd whisper.cpp
make
./models/download-ggml-model.sh base
./main -m models/ggml-base.bin -f samples/jfk.wav

🔊 Step 8: Install Piper (Text-to-Speech)

sudo apt install pip3
pip3 install piper-tts
piper --list-voices
piper --model en_US-lessac-medium.onnx --text "Hello there!"

🎉 You're Done!

You now have a complete local AI workstation, capable of handling offline LLMs, speech-to-text, automation, and voice synthesis — all on your own machine.