Learning_roadmap.md

AI Engineering Lab — Learning Roadmap

Goal

Build a practical AI engineering skillset around local LLMs, workflow automation, RAG, agents and deployable AI systems.

Phase 1 — LLM foundations

Project 01 — llm_playground

Focus:

• tokens;
• sampling;
• temperature;
• top-k;
• top-p;
• inference experiments.

Project 02 — local_llm_server

Focus:

• local model serving;
• OpenAI-compatible APIs;
• Ollama / llama.cpp / vLLM concepts;
• latency and throughput.

Phase 2 — Workflow automation

Project 03 — workflow_automation_lab

Focus:

• n8n;
• JavaScript Code Nodes;
• HTTP/API ingestion;
• scraping-light workflows;
• parsing;
• normalization;
• deduplication;
• scoring;
• storage;
• alerting.

First use case:

• BD Kids Hunter

This project is not RAG. It prepares the data-pipeline mindset needed for RAG and agents.

Phase 3 — RAG systems

Project 04 — rag_system

Focus:

• basic retrieval augmented generation;
• embeddings;
• vector database;
• chunking;
• document Q&A.

Project 05 — rag_enterprise

Focus:

• metadata filtering;
• permissions;
• incremental updates;
• citations;
• evaluation;
• production patterns.

Phase 4 — Agents

Project 06 — ai_agents_lab

Focus:

• tools;
• state;
• planning;
• LangGraph-style architecture;
• human-in-the-loop.

Project 07 — business_watch_agent

Focus:

• automated monitoring;
• source comparison;
• market watch;
• supplier watch;
• server hardware watch;
• alert generation.

Phase 5 — Vision and multimodal AI

Project 08 — vision_ai_lab

Focus:

• image analysis;
• visual classification;
• object detection foundations.

Project 09 — multimodal_assistant

Focus:

• text + image + audio workflows;
• assistant interaction patterns.

Phase 6 — Robotics and platform

Project 10 — robot_ai_system

Focus:

• robotics AI;
• perception;
• control;
• local inference.

Project 11 — ai_platform

Focus:

• integrated local AI platform;
• workflows;
• models;
• RAG;
• agents;
• monitoring.