Home › Courses › Build a Local AI Dev Environment
Build a Local AI Dev Environment
Local models, local chat, local RAG, local code assistant — zero cloud
1 phase. 6 lessons. 6 labs. Build a fully private AI stack on your own machine: Ollama running local models, Open WebUI as your local ChatGPT, Qdrant for local RAG over your own documents, and a local code assistant in VS Code — all orchestrated with one docker-compose up command.
6 lessons · 6 labs · 4 services · 1 private stack
Outcomes you will have by the end
- A fully private, running AI stack — Local chat, local RAG over your own documents with citations, and a local code assistant in VS Code — all running on your own hardware, provably (system monitor visible, zero network calls to any AI provider).
- A working mental model of Docker Compose networking — host.docker.internal for reaching the host machine, service names for reaching other containers on the same network — the exact distinction that trips up most multi-service Docker setups.
- A one-command, reproducible setup — docker-compose.yml + setup.sh + .env.example that rebuilds your entire stack on a new machine — tested with a real teardown/rebuild cycle, not just assumed to work.
- Verified certificate of completion — Issued by DeVenture Academy, tied to your completion record. Lists the local-LLM, Docker Compose, and RAG skills you built.
What you will be able to do
Local LLM Operations · Docker Compose Orchestration · Local RAG Architecture · IDE AI Integration · Infrastructure Reproducibility
Every phase, every lesson, every project
- Build a Local AI Dev Environment (6 lessons) — Hardware sizing and quantization, Ollama models via CLI and API, Open WebUI via Docker Compose, local RAG with Qdrant, a local code assistant in VS Code, one-command reproducibility
The technologies you will use
Ollama · Open WebUI · Qdrant · Docker Compose · Continue.dev
Roles this course prepares you for
- Privacy-First Developer (Cost savings — zero ongoing API spend) — Run your own AI tools entirely locally for sensitive work, personal documents, or offline use, with no per-token cost and no data leaving your machine.
- AI Infrastructure Consultant ($100-250/hr) — Help teams and individuals set up private, on-premise AI stacks for compliance, cost, or data-sovereignty reasons.
- Developer (any specialty) (Varies by employer) — Use a private code assistant and local chat tool as part of your daily workflow, independent of any vendor's API uptime or pricing changes.
Common questions
What hardware do I need?
8GB RAM is the floor (Q4 8B models run at roughly 5-10 tokens/sec), 16GB+ is comfortable. No dedicated GPU required, though one helps. Lesson 1 has you check your own hardware before choosing model sizes rather than guessing.
Is local AI as good as cloud APIs?
Honestly, no — a 7-8B local model is a real capability tier below frontier cloud models, and this course says so directly rather than overselling it. What you get instead: total privacy, zero ongoing API cost, and offline capability. The course teaches you where that tradeoff is worth it.
Why Qdrant instead of Open WebUI's default vector store?
Open WebUI now has first-class Qdrant support via a couple of environment variables — simpler than it used to be. Qdrant gives you a real, inspectable vector database (with its own dashboard) rather than an opaque built-in store, useful if you want to understand or extend the RAG pipeline later.
What do I end up with?
A complete local AI stack: Ollama running 3 models, Open WebUI as your chat interface, Qdrant powering RAG over your own documents, Continue.dev in VS Code for code assistance, and a one-command setup script that rebuilds all of it — verified with a real teardown/rebuild test.
Key terms in this course
RAG (Retrieval-Augmented Generation) · Orchestration · Quantization
Start the Build a Local AI Dev Environment course
Create a free account — the opening phases of 24 of 30 courses are free, no credit card. Or see Pro pricing.
All courses · Pricing · About · FAQ · Glossary