All learning paths
Level 3Agentic AI Engineering

Agentic AI Engineer Path

Build autonomous AI systems

Visual Roadmap

10 stages · click any stage to jump

L3 · Agentic AI Engineering
1

LLM Fundamentals

Function calling & tool schemas

2

Tool Calling

Dynamic execution & JSON schemas

3

Agent Loops

ReAct pattern & reasoning chains

4

State & Memory

Short-term buffers & persistent graphs

5

Planning

Plan-and-execute decomposition

6

Workflows

LangGraph state channels

7

Multi-Agent Systems

Supervisor-worker orchestration

8

MCP

Model Context Protocol client/servers

9

Agent Evaluation

Trajectory evaluation & mock tests

10

Production Agents

Guardrails, timeouts & sandboxing

StartCapstone

Overview

Master agentic AI: tool calling, agent loops, state and memory, planning, multi agent orchestration, MCP, and production agent systems with evaluation and security.

Learning outcomes

  • Design and build production AI agents
  • Implement agent loops with tool calling and memory
  • Orchestrate multi-agent systems and workflow graphs
  • Apply MCP for tool ecosystems
  • Enforce human-in-the-loop approvals and recursion guards

Stages

1

Function calling & tool schemas

Start Stage 1
2

Dynamic execution & JSON schemas

Start Stage 2
3

ReAct pattern & reasoning chains

Start Stage 3
4

Short-term buffers & persistent graphs

Start Stage 4
5

Plan-and-execute decomposition

Start Stage 5
6

LangGraph state channels

Start Stage 6
7

Supervisor-worker orchestration

Start Stage 7
8

Model Context Protocol client/servers

Start Stage 8
9

Trajectory evaluation & mock tests

Start Stage 9
10

Guardrails, timeouts & sandboxing

Start Stage 10

Series in this path