AI and machine learning · Free · ₹0
Private AI with Local LLMs: Ollama, Open WebUI and Offline RAG
Run AI on your own hardware so confidential drawings, specifications and plant data never leave the building: open-weight models and their licences, GPU and RAM sizing, quantisation, Ollama, Open WebUI for a team, the API and Python, offline RAG with citations, datasheet extraction, vision models, benchmarking, air-gapped deployment and an acceptable use policy.
Inside the course
From the lessons
Why run AI locally: confidentiality, export control, cost and offline sites
Matt Williams
Hardware sizing: VRAM, RAM, CPU-only and quantisation (GGUF, Q4, Q8)
Matt Williams
Open WebUI: a private ChatGPT for the team
Christian Lempa
Offline RAG over drawings, specifications and manuals
Matt Williams
Vision models locally: reading nameplates and P&ID snippets
Leon van Zyl
Air-gapped deployment, updates and access control
Matt Williams
Lesson frames belong to the creators named in the Credits below and are shown from YouTube.
What you will learn
Explain when a model must run locally and classify documents by where they may go; compare open-weight models and their licences; size VRAM, RAM and speed for a model, quantisation and context length; install and configure Ollama and write Modelfiles; run Open WebUI for a team with roles and groups; call models from the REST API and Python; build an offline RAG pipeline that cites and refuses; extract structured data from datasheets and nameplates with schema validation; benchmark local against hosted models on your own test set; deploy into an air-gapped network with checksums; and write the acceptable use policy that governs it.
- Decide which documents may go to a hosted AI tool, which must stay on a local model and which go nowhere
- Compare open-weight models and their licences: Apache 2.0, MIT and the Llama community licence
- Size GPU memory and speed for a model from its parameters, quantisation, context length and users
- Install and configure Ollama, and write a Modelfile your whole team can use
- Run Open WebUI as a private team chat server with roles, groups and model presets
- Build offline RAG over specifications that cites the clause and says NOT FOUND
- Extract datasheet and nameplate data with a JSON schema, validation and a human check
- Benchmark local against hosted models, deploy air-gapped with checksums and write the acceptable use policy
The course project · about 14 hours
A private team AI server that answers from project specifications with citations, sized, tested and governed
Stand up Ollama and Open WebUI as a team server, index a project's specification PDFs in a knowledge collection, make a model preset that cites the clause and says NOT FOUND, size the hardware for the team, test it on twenty questions, and write the acceptable use policy that lets engineers use it on confidential documents.
Sample document pack, 5 documents, filled in for the scenario
- URSRequirements for the private specification assistant
- SDSDesign and sizing note for the team AI server
- ProcedureAcceptable use policy for AI tools
- Test reportAcceptance test of the specification assistant on project 2607
- Risk registerRisk register for the office AI server
Read inside the course and download as a workbook. The project is optional practice, marked when you submit it; the certificate needs only the modules and the final assessment.
Course content
13 modules · 40 lessons · 10h 40m
In order, at whatever pace suits you. Each module ends with a practice task that builds on the last.
- 01Why run AI locally: confidentiality, export control, cost and offline sites4 lessons33m
- 02Open-weight models and their licences: Llama, Qwen, Gemma, Mistral and gpt-oss4 lessons31m
- 03Hardware sizing: VRAM, RAM, CPU-only and quantisation (GGUF, Q4, Q8)4 lessons57m
- 04Installing Ollama: the CLI, environment variables and the Modelfile5 lessons51m
- 05Open WebUI: a private ChatGPT for the team2 lessons1h 6m
- 06The Ollama REST API and Python3 lessons3h 17m
- 07Offline RAG over drawings, specifications and manuals4 lessons54m
- 08Structured output: extracting data from datasheets2 lessons22m
Requirements
- Who it is for
- Intermediate. For engineers in plants, EPC contractors, consultancies, defence and oil and gas who want AI on data they cannot paste into a hosted chatbot. You should be comfortable with a command line and able to run a short Python script; Generative AI and LLM Foundations is the natural course before this one.
- Software
- Ollama, Open WebUI (in Docker), Python with the ollama, chromadb and pydantic packages, and optionally LM Studio and promptfoo. All are free to download. What to download, and how
- Hardware
- A laptop with 16 GB of RAM runs the small models used in the modules. A GPU with 8 GB or more of memory, or an Apple silicon Mac with 16 GB or more, makes them faster. The sizing module shows how to size a team server; you do not need to buy one.
Software you need
What to download, where from, what it costs and how to install it. Every link goes to the maker's own site, never a mirror.
Open WebUI runs in Docker from ghcr.io/open-webui/open-webui, so it needs no separate download. On a Linux server use Docker Engine rather than Docker Desktop. LM Studio and promptfoo are optional.
Required
- 01Free
Ollama
Ollama Inc.
- Runs on
- macOS 14 Sonoma or later, Windows 10 or later, Linux
- Account
- None needed
Ollama is free and open source (MIT licence). Each model you download has its own licence; check it before commercial use.
Steps
- 1.Open ollama.com/download and pick your system.
- 2.On Windows run OllamaSetup.exe; on macOS open the DMG; on Linux run: curl -fsSL https://ollama.com/install.sh | sh
- 3.Open a terminal and run a small model, for example: ollama run <model name> (browse names on ollama.com).
- 4.Type a question at the prompt. Type /bye to exit.
- Models are several GB each. Check your free disk space first.
- More RAM, or a supported GPU, lets you run larger models faster.
- On Windows you can also install from PowerShell: irm https://ollama.com/install.ps1 | iex
Official download pageollama.com - 02Free
Python
Python Software Foundation
- Runs on
- Windows (not Windows 7 or earlier), macOS and Linux. Windows builds for x64, 32-bit and Arm64.
- Account
- None needed
Python is free, open source software. You can use it for learning and for commercial work at no cost.
Steps
- 1.Open python.org/downloads and click the download button for your system.
- 2.On Windows, run the Python install manager (or the classic 64-bit installer) you downloaded.
- 3.If you use the classic installer, tick "Add python.exe to PATH" on the first screen, then click Install Now.
- 4.Open a new Command Prompt or Terminal and run: python --version (on Windows you can also run: py --version).
- 5.Install packages with pip, for example: python -m pip install requests
- pip comes with Python. Run it as python -m pip so it always matches the Python you are using.
- Python.org now recommends the Python install manager on Windows. If it offers to add its folder to PATH, say yes so the python command works everywhere.
- Make a virtual environment for each project: python -m venv .venv
Official download pagepython.orgAlternatives
- Anaconda Distribution: Python with 600+ data science packages and Jupyter already included.
- 03Free for personal use, education and small businesses
Docker Desktop
Docker, Inc.
- Runs on
- Windows 10 64-bit version 22H2 (build 19045) or Windows 11 64-bit version 23H2 (build 22631) or later, with WSL 2 and hardware virtualisation on, 8 GB RAM. Also macOS and Linux.
- Account
- None needed
Free for personal use, education, non-commercial open source projects, and businesses with fewer than 250 employees and less than 10 million US dollars in annual revenue. Larger businesses need a paid subscription. Docker Engine on Linux is open source and is not covered by these terms.
Steps
- 1.Open docker.com/products/docker-desktop and download the installer for your system.
- 2.On Windows, install WSL first (wsl --install) if you do not have it.
- 3.Run the installer and keep the WSL 2 option selected.
- 4.Restart if asked, then start Docker Desktop and accept the subscription terms.
- 5.Open a terminal and run: docker run hello-world
- Turn on virtualisation in BIOS or UEFI if Docker says it is not available.
- On Linux servers, install Docker Engine instead of Docker Desktop.
Official download pagedocker.comAlternatives
- Docker Engine (Linux): Free, open source engine for Linux. Not covered by the Docker Desktop subscription terms.
Optional
Useful, not needed to finish the course.
- 04Free
Visual Studio Code
Microsoft
- Runs on
- Windows 64-bit (supported Windows client versions), macOS (latest and two previous releases), Linux (Ubuntu 20.04, Debian 10, RHEL 8, Fedora 36 or later)
- Account
- None needed
- Size
- Less than 200 MB download, under 500 MB installed
Free to download and use. Extensions from the Marketplace each have their own licence.
Steps
- 1.Open code.visualstudio.com/download.
- 2.Pick the installer for your system (on Windows, the User Installer is fine).
- 3.Run the installer. On Windows, tick "Add to PATH" if it is offered.
- 4.Open VS Code and install the extensions your course uses, for example the Python extension.
- Hardware needs are small: a 1.6 GHz processor and 1 GB of RAM.
- On Windows, the "Open with Code" options in the installer let you open folders from File Explorer.
Official download pagecode.visualstudio.com
Checked against each maker's own page on 27 September 2026. Trial lengths and editions change; the maker's page is the final word.
Private AI with Local LLMs: Ollama, Open WebUI and Offline RAG at a glance
Private AI with Local LLMs: Ollama, Open WebUI and Offline RAG is a free, self-paced online course from EDWartens for engineers at plants, EPC contractors, consultancies, defence and oil and gas operators in India, the Gulf and worldwide who need AI on confidential documents, and the IT and OT staff who run the server. It has 13 modules and 10h 40m of video lessons by Matt Williams, IBM Technology, freeCodeCamp.org and others, with written notes and worked problems, a practical project with a document pack and a 15-question final assessment (pass mark 60%). Learning is free with an account; an optional certificate with a public verification code is issued when you pass. Last updated 27 September 2026.
All course factsHide course facts
- Price
- ₹0, free for good. No trial, no card. Comparable classroom training of this length costs about ₹6,999.
- Who it is for
- Engineers at plants, EPC contractors, consultancies, defence and oil and gas operators in India, the Gulf and worldwide who need AI on confidential documents, and the IT and OT staff who run the server
- Format
- 13 self-paced modules, 10h 40m of video, written notes, a practice task per module and one final assessment.
- Level
- Intermediate. Intermediate. For engineers in plants, EPC contractors, consultancies, defence and oil and gas who want AI on data they cannot paste into a hosted chatbot. You should be comfortable with a command line and able to run a short Python script; Generative AI and LLM Foundations is the natural course before this one.
- Brand
- Vendor-neutral
- Software
- Ollama, Open WebUI (in Docker), Python with the ollama, chromadb and pydantic packages, and optionally LM Studio and promptfoo. All are free to download.
- Hardware
- A laptop with 16 GB of RAM runs the small models used in the modules. A GPU with 8 GB or more of memory, or an Apple silicon Mac with 16 GB or more, makes them faster. The sizing module shows how to size a team server; you do not need to buy one.
- Certificate
- Optional EDWartens Certificate of Completion, verifiable by code. Not a vendor credential.
- Video lessons by
- Matt Williams, IBM Technology, freeCodeCamp.org, Christian Lempa, NetworkChuck, Tech With Tim, Syntax, Real Python, Sam Witteveen, pixegami, Case Done by AI, Professor Patterns, Tony Kipkemboi, Ian Wootten, Leon van Zyl, Skill Leap AI, Jason Koo, BlueSpork, AppSecEngineer (independent creators, credited below)
- Language
- English
- Last updated
- 27 September 2026
A shareable EDWartens certificate
Finish every module and pass the final assessment, and the optional EDWartens certificate is yours. It carries a unique verification code on a public page anyone can check, so it stands up when a recruiter looks it up. See it below.
The course itself stays free whether or not you ever buy one.
Stuck? Ask a practising engineer
A free course usually means a comment section and hope. This one does not. Every module has an Ask-your-trainer panel that reaches the same engineers who teach our paid programme: people who commission panels for a living, not moderators.
Pairs well with
PLC programming · FreeSiemens TIA PortalFrom zero electrical knowledge to a working, simulated S7-1200 program, for nothing.
PLC programming · FreeSiemens TIA Portal in Three HoursThe first three hours of the Siemens TIA Portal course, cut to end on a win: what a PLC is, how it is wired, a project configured in TIA Portal, and your first ladder program running in simulation. Finish it in an evening or two, earn a certificate, and carry straight on into the full course.
PLC programming · FreeTIA Portal: Build a MachineOne machine, start to finish. Take a bottle filling line from a written specification and an I/O list to a structured S7-1200 program with a fill station, a capper, a reject sorter and an operator screen with alarms, then test it against a written record and archive it for hand-over. The lessons are the reference; the machine is yours, and it is what you submit.
Instrumentation · FreeInstrumentation for PLC EngineersThe half of the loop that is not code. Follow one measurement from the transmitter in the field, down the 4-20 mA loop, into the analog card, through NORM_X and SCALE_X into engineering units, out again to a valve, and back to the control room when the reading is wrong.More free courses: Free AI courses for engineers · Free generative AI and ChatGPT courses
Learner reviews
No reviews yet
Reviews here are written only by learners who have finished every module of Private AI with Local LLMs: Ollama, Open WebUI and Offline RAG, and they are published exactly as written. Finish the course and yours will be the first.
Common questions
Who is this private AI and local LLM course for?
It is for engineers who want to use AI on documents they cannot paste into a hosted chatbot: process, instrument, electrical and project engineers at EPC contractors, consultancies, plants, defence and PSU projects, and oil and gas operators in India, the Gulf and elsewhere. IT and OT staff who will run the server for them will find it useful too.
Is the course free to learn?
Yes. Every module, the notes, the worked problems, the project and the final assessment are free, and every piece of software used is free to download. The certificate is optional.
What do I need to know first?
You should be comfortable with a command line and able to run a short Python script. If tokens, context windows and RAG are new to you, take Generative AI and LLM Foundations first; after this course, RAG and Chatbots with LangChain goes deeper into retrieval.
What hardware do I need to run local LLMs?
A laptop with 16 GB of RAM is enough for the small models used in the modules. A GPU with 8 GB or more of memory, or an Apple silicon Mac with 16 GB or more, makes them several times faster. Module 3 teaches you to size a team server from the model, context length and number of users, so you can write the purchase request yourself.
Which software does it use?
Ollama to run the models, Open WebUI in Docker as the team chat interface, and Python with the ollama, chromadb and pydantic packages for the API, offline RAG and extraction modules. LM Studio and promptfoo appear as optional tools. Versions move every few weeks, so the notes say what was current in October 2026 and tell you to check.
Is a local model safe to use on confidential and export-controlled data?
A local model keeps the data on hardware you control, which is what most confidentiality clauses require, but it does not make the answers correct. The course teaches the data classes, the locked-down server, the logging and the acceptable use policy; export-controlled data still needs your security officer's approval for any tool, and no AI output reaches live plant without review and test.
How long does the Private AI with Local LLMs course take?
About 18 hours of video, notes and practice at your own pace, of which about 11 hours is video. The optional team server project takes about 14 hours more.
What certificate does the Private AI with Local LLMs course give?
An EDWartens certificate of completion, issued when you finish the modules and pass the 15-question final at 60 percent, with a number anyone can verify on our site. It is not a certification from Ollama, Open WebUI or any model maker, none of whom is affiliated with EDWartens.
Is the Private AI with Local LLMs course free in India, and what does the certificate cost?
Yes. Learning costs ₹0 in India: every module, the written notes, the practice tasks and the final assessment, with no card and no trial period. The only paid item is the optional EDWartens Certificate of Completion, ₹559 including GST for this intermediate course, paid in rupees through Razorpay, and only if you want it after passing the final assessment.
What you walk away with
Your certificate for Private AI with Local LLMs: Ollama, Open WebUI and Offline RAG
Finish the course, pass the final, and this is the document with your name on it.
Verifiable by anyone
Adds to LinkedIn in one click
QR code on the certificate
Names what you can do
A permanent link
Earned, not attended
Learning is free. The certificate is optional.
Add it now and pay only when you have finished the course, or come back for it later. One-off, US$28.99, with a receipt.
Issued by EDWartens India (Wartens Automation Private Limited) as a Certificate of Completion for this self-paced course. It is not a vendor certification, a university award or a CPD-accredited activity, and it does not certify competence on live equipment. Delivered electronically; see the refund policy.
Credits
Who made the video lessons
The video lessons in this course were created by the people below, not by EDWartens. Every lesson streams from its creator's own YouTube channel; EDWartens neither hosts nor sells that footage, and the creators are not affiliated with EDWartens and do not endorse this course. What EDWartens wrote is the study plan, the notes, the practice tasks and the assessments.
- Matt Williamsthe free Ollama Course lessons on installing Ollama, the CLI, environment variables, custom models, model types, quantisation, the file layout, the OpenAI-compatible API, RAG and embeddings
- IBM Technologythe explainers on Ollama, open-source LLMs, privacy-preserving AI at home, open against closed AI, Llama, llama.cpp, LLM benchmarks, LLM as a judge, AI governance and AI data exposure, and the local deployment recap in the final module
- freeCodeCamp.orgthe three-hour Ollama course by Paulo Dichone, building local AI apps with the API and Python
- Christian Lempathe self-hosted Ollama and Open WebUI platform build
- NetworkChuckhosting the full local AI stack on a Linux GPU server with per-user restrictions
- Tech With Timthe LM Studio tutorial
- Syntaxthe episode on local AI hardware, setup and models
- Real Pythonrunning models locally with Ollama and Python
- Sam Witteveenbuilding a vision app with Ollama structured outputs
- pixegamithe Python RAG tutorial with local LLMs over PDFs
- Case Done by AIthe complete guide to Open WebUI admin settings
- Professor PatternsOpen WebUI knowledge bases, RAG and embeddings
- Tony Kipkemboistructured outputs with Ollama and Pydantic models
- Ian WoottenOllama structured outputs in five minutes
- Leon van ZylLlama 3.2 Vision with Ollama, chatting with images locally
- Skill Leap AIrunning OpenAI's open-weight gpt-oss models locally
- Jason Koogetting started with promptfoo for testing models
- BlueSporkrunning a model fully offline from a USB drive
- AppSecEngineerthe guide to the OWASP Top 10 for LLM applications
If you are one of these creators and would like a lesson removed or credited differently, write to info@wartens.com.