local-ai-environment

Local AI Environment

Modular PowerShell toolkit to prepare a Windows laptop as a local AI development environment.

Architecture

Windows
  └── Open-WebUI (port 3000)   ← browser interface + API gateway
        └── Ollama (port 11434) ← local model runtime
              └── LLM model     ← qwen2.5, phi4, llama3.x, gemma3, …

  (separate, prepared for future use)
  └── Docker
        └── n8n (port 5678)    ← Phase 2: local automation platform

Open-WebUI and Ollama are installed natively (no Docker). Docker is audited/cleaned separately to be ready for the n8n phase.

Prerequisites

Requirement	Notes
Windows 10/11 (64-bit)
PowerShell 5.1+	Built into Windows
Python 3.11+	Required for Open-WebUI (`pip install open-webui`)
Internet access	Required during installation only
~10 GB free disk	Minimum for a small model

Quick Start — Execution Order

Run from the project root in a PowerShell session (no elevated privileges required for most steps; Ollama installer may prompt for admin):

# 1. Audit your system
.\scripts\00-System-Check.ps1

# 2. Audit existing Docker environment (safe, no changes)
.\scripts\01-Docker-Audit.ps1

# 3. (Optional) Clean Docker for future n8n use
.\scripts\02-Docker-Clean.ps1 -Mode SAFE

# 4. Install Open-WebUI
.\scripts\03-Install-OpenWebUI.ps1

# 5. Install Ollama
.\scripts\04-Install-Ollama.ps1

# 6. Download and configure a local LLM model
.\scripts\05-Configure-Local-LLM.ps1

# 7. Connect Open-WebUI to Ollama
.\scripts\06-Configure-OpenWebUI.ps1

# 8. Validate the full environment
.\scripts\07-Validate-Environment.ps1

# 9. Generate final summary report
.\scripts\99-Generate-Report.ps1

Script Modes

Most scripts support these common parameters:

Parameter	Description
`-DryRun`	Show what would be done; make no changes
`-NonInteractive`	Skip all confirmation prompts (use safe defaults)
`-Verbose`	Extra diagnostic output

02-Docker-Clean.ps1 also accepts -Mode AUDIT|SAFE|FULL.

Directory Structure

Local-AI-Environment/
├── scripts/        Numbered PowerShell scripts (execution order)
├── modules/        Shared PowerShell modules (Logging, System, Docker, …)
├── config/         environment.psd1 and models.psd1 settings
├── docs/           Architecture, installation and operations guides
├── logs/           Runtime log files (git-ignored)
├── reports/        Generated JSON/MD reports (git-ignored)
└── tests/          Future test scripts

Documentation

See docs/ for detailed guides:

ARCHITECTURE.md — design decisions and component overview
INSTALLATION.md — step-by-step installation guide
OPERATIONS.md — daily usage
TROUBLESHOOTING.md — common problems and fixes
ROADMAP.md — planned future phases
DOCKER-N8N-PLAN.md — n8n integration architecture

This project is documented as part of a hands-on article series on building a local AI stack:

Substack (PL): pawellach.substack.com — Lokalne środowisko AI series
Medium (EN): medium.com/@pawellach — Local AI Infrastructure series

The series covers Ollama setup, Open-WebUI configuration, RAG pipelines, and MCP server integration — all with this repository as the reference implementation.

License

MIT. Private use only — no warranty.