Personal AI engineering learning lab.
This repository organizes a 2026 learning path toward practical AI engineering, with a strong focus on:
- local LLMs;
- workflow automation;
- RAG;
- agents;
- business watch systems;
- multimodal AI;
- robotics;
- deployable AI platforms.
| Order | Project | Focus |
|---|---|---|
| 01 | 2026_project_01--llm_playground |
LLM sampling, inference parameters, basic experiments |
| 02 | 2026_project_02--local_llm_server |
Local LLM API server, OpenAI-compatible patterns |
| 03 | 2026_project_03--workflow_automation_lab |
n8n, JS Code Nodes, ingestion, parsing, normalization, scoring, alerts |
| 04 | 2026_project_04--rag_system |
Basic Retrieval Augmented Generation |
| 05 | 2026_project_05--rag_enterprise |
Production-grade RAG patterns |
| 06 | 2026_project_06--ai_agents_lab |
Agents, tools, state, human-in-the-loop |
| 07 | 2026_project_07--business_watch_agent |
Watch systems combining workflows, memory, retrieval and agents |
| 08 | 2026_project_08--vision_ai_lab |
Computer vision foundations |
| 09 | 2026_project_09--multimodal_assistant |
Multimodal assistant experiments |
| 10 | 2026_project_10--robot_ai_system |
Robotics AI experiments |
| 11 | 2026_project_11--ai_platform |
Integrated local AI platform |
LLM foundations
↓
workflow automation
↓
RAG systems
↓
agents
↓
business watch
↓
vision / multimodal / robotics
↓
AI platform
RAG and agents require solid data-pipeline thinking.
Before building retrieval systems, it is useful to learn how to:
- fetch data;
- parse sources;
- normalize records;
- deduplicate;
- score;
- store;
- alert;
- manage uncertainty.
This is the role of 2026_project_03--workflow_automation_lab.
Its first use case is BD Kids Hunter.
Learning_roadmap.mdTiming_table.mddocs/project_taxonomy.md