public

Open Source AI Ecosystem

From local inference to production serving — the complete open-source AI stack.
Co-Created by Kiran Shirol and Claude
Core Stack Hugging Face Ollama vLLM llama.cpp LoRA
home Learning Portal play_arrow Start Learning 15 chapters · One comprehensive chapter per topic
Foundation

The Open Source Stack

What the ecosystem is, the Hugging Face universe, open models, and why open source is winning.
1
lightbulb
What Is the Open Source AI Ecosystem
The closed vs open spectrum, why open source matters, and the 4 layers of the stack.
arrow_forward Learn
2
hub
The Hugging Face Hub
Models, datasets, Spaces, model cards, leaderboards, and license-aware selection.
arrow_forward Learn
3
library_books
The Hugging Face Libraries
Transformers, Diffusers, PEFT, TRL, Accelerate, Datasets — what each does.
arrow_forward Learn
4
category
The Open Source Model Landscape
Llama, Mistral, Gemma, Qwen, DeepSeek, Phi — benchmark signals and license constraints.
arrow_forward Learn
Local AI

Running Models on Your Hardware

GGUF quantization, Ollama, llama.cpp, and GUI tools for local AI.
5
compress
GGUF & Quantization
FP32 to INT4, size vs quality tradeoffs, and the GGUF format explained.
arrow_forward Learn
6
terminal
Running LLMs Locally with Ollama
One-command setup, CLI, REST API, Modelfile, and the llama.cpp backend.
arrow_forward Learn
7
memory
llama.cpp — The Inference Engine
C/C++ speed, hardware coverage from H100 to Raspberry Pi, GGUF conversion.
arrow_forward Learn
8
desktop_windows
GUI Tools — LM Studio & Open WebUI
Desktop apps and self-hosted interfaces for non-CLI AI workflows.
arrow_forward Learn
Fine-Tuning

Adapting Models to Your Needs

LoRA, QLoRA, and the tools that make fine-tuning accessible on consumer hardware.
9
tune
LoRA & QLoRA — Parameter-Efficient Fine-Tuning
Rank decomposition, adapter training, and practical fine-tuning on consumer hardware.
arrow_forward Learn
10
build
Fine-Tuning Toolkits — Unsloth, Axolotl & LLaMA-Factory
The tools that make LoRA/QLoRA fast, flexible, and accessible.
arrow_forward Learn
Production

Deploying at Scale

High-throughput inference engines and how to choose between them.
11
speed
vLLM & PagedAttention
How PagedAttention improves KV-cache efficiency and unlocks higher serving throughput.
arrow_forward Learn
12
compare
Inference Engines Compared
TGI, TensorRT-LLM, SGLang, and vLLM — decision matrix for every workload.
arrow_forward Learn
Applications

Building with the Open Stack

Orchestration frameworks, end-to-end architecture, and the open-source AI community.
13
account_tree
LangChain, LlamaIndex & DSPy
Orchestration, RAG pipelines, and programming language models — when to use each.
arrow_forward Learn
14
layers
The Open Source AI Stack — End-to-End
Full architecture from model download to production app, with cost comparison.
arrow_forward Learn
15
diversity_3
Contributing to Open Source AI
HF Hub uploads, llama.cpp PRs, licensing landscape, and the OSS AI community.
arrow_forward Learn
explore

Explore Related Courses

psychology
Agentic AI
Planning, Memory & Tool Use
bolt
AI Infrastructure
GPUs, Distributed Training & Serving
tune
Fine-Tuning
LoRA, QLoRA, RLHF & Adapters