# Where to Find Structured Learning Paths in the AI Engineering From Scratch Curriculum

> Find structured AI Engineering From Scratch learning paths in the rohitg00/ai-engineering-from-scratch repo. Access JSON roadmap files in the learning-paths directory for a clear learning journey.

- Repository: [Rohit Ghumare/ai-engineering-from-scratch](https://github.com/rohitg00/ai-engineering-from-scratch)
- Tags: getting-started
- Published: 2026-08-28

---

**The structured learning paths for the AI Engineering From Scratch curriculum are defined as JSON roadmap files located in the `learning-paths/` directory of the repository.**

The `rohitg00/ai-engineering-from-scratch` repository organizes its curriculum through machine-readable JSON manifests that define complete skill tracks. These **structured learning paths** specify prerequisites, lesson sequences, and estimated completion times for each engineering track. All path definitions follow a consistent schema and link to concrete lesson implementations under the `phases/` directory.

## Locating the Structured Learning Path Definitions

All curriculum roadmaps reside in the **`learning-paths/`** directory at the repository root. Each JSON file represents a distinct skill track with its own prerequisites and ordered lessons.

The repository currently maintains the following **structured learning paths**:

- **[`learning-paths/agent-skills.json`](https://github.com/rohitg00/ai-engineering-from-scratch/blob/main/learning-paths/agent-skills.json)** — Defines the **Agent Skills Engineering** track, including prerequisites and lesson sequences such as "22-skills-and-agent-sdks" and "24-skill-discovery-and-progressive-disclosure"
- **[`learning-paths/model-context-protocol.json`](https://github.com/rohitg00/ai-engineering-from-scratch/blob/main/learning-paths/model-context-protocol.json)** — Describes the **Model Context Protocol (MCP)** track, organized into core, bidirectional, secure, and advanced lesson groups
- **`learning-paths/`** — The container directory for all learning path manifests, with additional tracks added as the curriculum expands

## JSON Schema and Path Structure

Each learning path file follows a standardized schema containing metadata fields and lesson arrays. According to the source code, the schema includes: `schemaVersion`, `id`, `title`, `summary`, `estimatedMinutes`, `prerequisites`, `lessons`, and `optionalLessons`.

Every lesson entry within the `lessons` array points to a concrete implementation folder under **`phases/`**. For example, a lesson with path `phases/13-tools-and-protocols/22-skills-and-agent-sdks` contains the lesson documentation, code examples, tests, and quizzes.

The directory structure follows this pattern:

```

phases/
└── 13-tools-and-protocols/
    └── 22-skills-and-agent-sdks/
        ├── docs/
        │   └── en.md          # Human-readable lesson content

        ├── code/
        │   └── main.*         # Runnable implementation

        └── tests/             # Validation tests

```

## Programmatic Access to Learning Paths

You can consume these structured learning paths programmatically to build custom study plans or validate curriculum completion.

### Reading a Specific Track with Python

Use the following Python script to fetch and parse the Agent Skills learning path directly from the repository:

```python
import json, pathlib, urllib.request

# Load the Agent Skills learning path from the repository

url = "https://raw.githubusercontent.com/rohitg00/ai-engineering-from-scratch/main/learning-paths/agent-skills.json"
data = json.loads(urllib.request.urlopen(url).read().decode())
print(f"Track: {data['title']}")
print("Lessons:")
for lesson in data["lessons"]:
    print(f"  • {lesson['order']}: {lesson['title']} ({lesson['path']})")

```

### Iterating Over All Available Paths

To list all available learning tracks locally, use this bash command with `jq`:

```bash

# List all JSON manifests in the learning-paths folder

for f in learning-paths/*.json; do
    echo "=== $f ==="
    jq '.title, .lessons[].title' "$f"
done

```

## Key Files in the Curriculum Structure

Understanding the relationship between path definitions and content helps navigate the curriculum efficiently:

| File | Role |
|------|------|
| [`learning-paths/agent-skills.json`](https://github.com/rohitg00/ai-engineering-from-scratch/blob/main/learning-paths/agent-skills.json) | Master definition for the Agent Skills Engineering learning route |
| [`learning-paths/model-context-protocol.json`](https://github.com/rohitg00/ai-engineering-from-scratch/blob/main/learning-paths/model-context-protocol.json) | Master definition for the Model Context Protocol (MCP) learning route |
| `phases/…/docs/en.md` | Human-readable lesson content referenced by each path entry |
| `phases/…/code/main.*` | Runnable implementation files for hands-on practice |
| `phases/…/tests/` | Unit tests that validate lesson code correctness |
| [`README.md`](https://github.com/rohitg00/ai-engineering-from-scratch/blob/main/README.md) & [`ROADMAP.md`](https://github.com/rohitg00/ai-engineering-from-scratch/blob/main/ROADMAP.md) | High-level overview documents linking phases and lessons |

## Summary

- **Structured learning paths** are stored as JSON files in the `learning-paths/` directory of `rohitg00/ai-engineering-from-scratch`
- The curriculum currently includes **Agent Skills Engineering** and **Model Context Protocol** tracks, defined in [`agent-skills.json`](https://github.com/rohitg00/ai-engineering-from-scratch/blob/main/agent-skills.json) and [`model-context-protocol.json`](https://github.com/rohitg00/ai-engineering-from-scratch/blob/main/model-context-protocol.json)
- Each JSON file follows a consistent schema with fields for `prerequisites`, `estimatedMinutes`, and ordered `lessons` arrays
- Lesson entries link to concrete folders under `phases/` containing documentation, code, and tests
- You can programmatically access these paths using standard HTTP requests and JSON parsers

## Frequently Asked Questions

### What file format are the structured learning paths stored in?

The structured learning paths are stored as **JSON files** with a specific schema that includes fields like `schemaVersion`, `id`, `title`, `summary`, `estimatedMinutes`, `prerequisites`, `lessons`, and `optionalLessons`. This machine-readable format allows for programmatic curriculum navigation and custom tooling integration.

### How do I navigate from a learning path definition to the actual lesson content?

Each lesson object in the JSON arrays contains a `path` property that maps to a directory under `phases/`. For example, if a lesson specifies `"path": "phases/13-tools-and-protocols/22-skills-and-agent-sdks"`, the actual content resides in that folder with [`docs/en.md`](https://github.com/rohitg00/ai-engineering-from-scratch/blob/main/docs/en.md) for theory, `code/` for implementations, and `tests/` for validation.

### Which learning tracks are currently available in the repository?

As of the latest commit, the repository includes two primary **structured learning paths**: the **Agent Skills Engineering** track (defined in [`learning-paths/agent-skills.json`](https://github.com/rohitg00/ai-engineering-from-scratch/blob/main/learning-paths/agent-skills.json)) and the **Model Context Protocol (MCP)** track (defined in [`learning-paths/model-context-protocol.json`](https://github.com/rohitg00/ai-engineering-from-scratch/blob/main/learning-paths/model-context-protocol.json)). The `learning-paths/` directory serves as the container for all current and future curriculum tracks.

### How are prerequisites handled in the curriculum structure?

Each learning path JSON file includes a `prerequisites` field at the track level, and individual lessons follow a sequential order defined by the `order` property in the `lessons` array. The schema supports `optionalLessons` for supplementary content that does not block progression through the core curriculum.