You will be able to
- Install Ollama and run open-source models such as Llama 3.2 from the command line
- Customize a local model with a Modelfile that sets a system prompt and temperature
- Call the Ollama REST API and the Python ollama library for chat, generation, and streaming
- Build a retrieval augmented generation (RAG) pipeline over a PDF using embeddings, a vector store, and LangChain
- Structure a multi-agent application in which specialized agents hand off work to an orchestrator
Good to know first
- Basic Python programming, including functions, dictionaries, and classes
- Comfort with the terminal and running shell commands
- Ability to create a virtual environment and install packages with pip
- Basic understanding of what a large language model is
Syllabus
14 lessons and a final project
- 1
What Ollama is and why run LLMs locally Free
Ollama is a free, open-source tool that downloads and runs large language models on your own computer through one consistent set of commands. Running models locally gives you privacy, lower cost, lower latency and freedom to switch and customise models.
- 2
Installing Ollama and your first chat with Llama 3.2 Beyond Pass
You will install Ollama on your computer, download the Llama 3.2 model, and chat with it in the terminal. You will also learn the slash commands that control the chat session.
3 code6 build steps2 exercises4 questions - 3
Browsing the model library, tags, and parameter sizes Beyond Pass
You learn how to find models in the Ollama library, read a model's tags and sizes, and understand what parameters, context length, embedding length and quantization mean. You also practise the terminal commands to run, list, inspect, and remove models.
1 code6 build steps3 exercises3 questions - 4
Managing models and vision with Gemma and LLaVA Beyond Pass
You learn to pull, list, run and remove local models with the ollama command, and to use LLaVA, a multimodal model, to describe an image and answer follow-up questions. You also see that a running model keeps the conversation history, so later questions can refer back to earlier ones.
3 code7 build steps3 exercises3 questions - 5
Creating a custom model with a Modelfile Beyond Pass
You will learn how a Modelfile customises an existing Ollama model with a base model, a parameter such as temperature, and a system prompt. You will then use ollama create to build your own named model and chat with it.
2 code4 build steps2 exercises3 questions - 6
Using the Ollama REST API: generate, chat, and JSON output Beyond Pass
You will learn how Ollama runs a local server at localhost:11434 that any program can talk to over HTTP. You will call the generate and chat endpoints with curl, switch off streaming, and request JSON output.
2 code4 build steps3 exercises4 questions - 7
Chatting in the Ollama desktop app and attaching a PDF Beyond Pass
You will learn how to chat with a local model in a desktop app, switch models, and pick a vision model for images. You will also see how a knowledge stack (a simple RAG setup) lets the model answer questions about your own PDF.
2 exercises - 8
Calling Ollama from Python: requests, the ollama library, and streaming Beyond Pass
You will call a local Ollama model from Python twice: first by sending raw HTTP requests to the REST API with the requests library, then with the ollama Python library, which wraps those same endpoints. You will also learn how to stream a reply so words appear as they are generated.
4 code6 build steps3 exercises3 questions - 9
Project: Grocery List Categorizer Beyond Pass
You will build a small Python program that reads a plain grocery list, asks a local Llama 3.2 model to sort it into categories, and saves the result to a new text file. You will also see how a clear prompt steers the model and how to swap models later.
1 code5 build steps3 exercises3 questions - 10
RAG concepts: chunks, embeddings, and vector stores Beyond Pass
This lesson explains how Retrieval Augmented Generation (RAG) lets a local model answer questions from your own documents. You will learn the indexing pipeline (load, split into chunks, embed, store) and the query pipeline (embed the question, retrieve similar chunks, pass them to the LLM).
2 exercises4 questions - 11
Building a RAG pipeline with LangChain and a vector store Beyond Pass
You will build a complete Retrieval-Augmented Generation (RAG) pipeline: load a PDF, split it into overlapping chunks, turn the chunks into embeddings stored in a Chroma vector database, and answer questions with a local Llama 3.2 model. You will also see how the query prompt, retriever and answer prompt link together in one chain.
6 code6 build steps3 exercises4 questions - 12
Multi-agent recruitment app: architecture overview Beyond Pass
This lesson shows what the AI Recruiter Agency app does and how it is organised: several AI agents, each a specialist, pass work to one another, all powered by a local Llama 3.2 model through Ollama. You will see the user flow from uploading a resume to receiving a full recommendation report, and why the code keeps every agent connected to the same model.
2 exercises3 questions - 13
Building the agents, matcher, and orchestrator Beyond Pass
This lesson shows how a set of local AI agents share one parent class, BaseAgent, which talks to a local Ollama model through an OpenAI-style client. You will see how each specialist agent (extractor, matcher, profile enhancer, recommender, screener) is built the same way, with a name, instructions, and a run method.
4 code5 build steps3 exercises3 questions - 14
Running the recruitment agency app and wrap-up Beyond Pass
You will see how a multi-agent recruitment app is launched with Streamlit and how an orchestrator passes a resume through specialised agents to produce analysis, job matches, screening and a recommendation. You will also see the course wrap-up and what the local Ollama approach makes possible.
1 code5 build steps2 exercises3 questions
Final project
Local Study Buddy: a custom model, RAG over your PDF, and a two-agent hand-off
Build a fully local assistant that answers questions about a PDF of your choice. It uses a custom Ollama model, a RAG pipeline and two cooperating agents. Nothing leaves your computer, and you will practise every skill from the course.
Its 7 tasks open with Beyond Pass.
Free preview · Lesson 1 in full
What Ollama is and why run LLMs locally
Ollama is a free, open-source tool that downloads and runs large language models on your own computer through one consistent set of commands. Running models locally gives you privacy, lower cost, lower latency and freedom to switch and customise models.
Words to know
- Large language model (LLM)
- An AI model trained on huge amounts of text that can understand and generate language.
- CLI (command line interface)
- A way to control a program by typing commands in a terminal instead of clicking buttons.
- Embedding
- A list of numbers (a vector) that represents the meaning of a piece of text.
- RAG (retrieval augmented generation)
- A method where relevant document chunks are found first and given to a language model so it answers using them.
- Latency
- The delay between sending a request and getting the response.
Lessons 2–14 and the final project
They open with Beyond Pass: every lesson of every course, in full. One payment: $14.99 for 90 days or $29 for a year. No subscription.
Made by Beyond from the video “Ollama Course – Build AI Apps Locally” by freeCodeCamp.org. The video plays in YouTube's own player. Watch it on YouTube · All lessons