Open Models in Practice: Hugging Face, Ollama, llama.cpp & vLLM
Coming soon BeginnerIntermediateAdvanced
Get notified on YouTube Udemy certificate: coming soon
- Udemy
- Coming soon
- Lessons
- 9
- Planned length
- About 2 h 44 min
- Projects
- 1
- Practicals
- Planned: a hands-on practical and a Udemy assignment in every section, plus a capstone
What you will be able to do
- Choose an open model and check its licence
- Chat with a local model and call it from Python
- Run a Hub model locally and via a hosted API
- Run a quantized model on CPU and size models to your hardware
- Serve an OpenAI-compatible endpoint on a GPU and pick an engine
- Pick and run a small model that fits your device
Syllabus
Working With LLM APIs: Open and Local Models
- Open-Weight Models Explained: Llama, Qwen, Gemma, Mistral, DeepSeek Choose an open model and check its licence
- Run AI Locally With Ollama (and LM Studio) Chat with a local model and call it from Python
- Hugging Face for Beginners: Hub, Pipelines, Inference Providers Run a Hub model locally and via a hosted API
- llama.cpp, GGUF and How Big a Model You Can Run Run a quantized model on CPU and size models to your hardware
- Serve Models at Scale: vLLM (and SGLang, TGI) Serve an OpenAI-compatible endpoint on a GPU and pick an engine
Frontier and Advanced: Reasoning Models and Inference
- Small and On-Device Models: Laptops, Phones and Browsers Pick and run a small model that fits your device
- Inference Optimisation: KV Cache and Prefix Caching Explain and measure what each optimisation saves
- Speculative Decoding and Continuous Batching Turn on speculative decoding and batching and measure the speed-up
- Project: Benchmark Your Model Serving Stack Compare two serving setups on your own GPU budget (verify price)
Take before
Lesson list from the ADITEJUS curriculum. Planned lengths are estimates per lesson; Udemy shows the final length.