Home Original page

GitHub - JonathanLaidler/API: Build a REST API hands-on: FastAPI, ML models, testing, deployment, and LLM API wrapping

title REST-API-Builder
emoji 🏥
colorFrom blue
colorTo green
sdk docker
pinned false

REST-API-Builder

A self-paced, hands-on tutorial that teaches you how to build, test, and deploy a REST API that serves a machine learning model, using fictious healthcare examples throughout.

By the end, you will have a working API that classifies clinical notes as urgent or routine, tested with pytest, deployed to the cloud, and powered by an LLM for explanations.

Who Is This For?

Researchers and data scientists who:

  • Have basic Python knowledge
  • Have completed (or are familiar with) the ci-cd-template tutorial
  • Want to learn how to serve ML models as REST APIs
  • Are new to web APIs, HTTP, and deployment

What You'll Build

A Clinical Urgency Prediction API that:

  1. Accepts a clinical note (free text)
  2. Classifies it as urgent or routine
  3. Returns the prediction with a confidence score
  4. Stores predictions for later retrieval
  5. Is deployed and accessible via the internet
  6. Uses an LLM to explain why a note is urgent or routine
flowchart LR
    A[Clinical Note] -->|POST /v1/predictions| B[FastAPI Server]
    B --> C[TF-IDF + LogReg Model]
    C --> B
    B -->|201 Created| D["{ prediction: urgent,\n confidence: 0.94 }"]

    style A fill:#e8f4f8,stroke:#2980b9
    style B fill:#fef9e7,stroke:#f39c12
    style C fill:#fadbd8,stroke:#e74c3c
    style D fill:#d5f5e3,stroke:#27ae60
Loading

Why a REST API (and Not Just pip install)?

pip install (library) REST API (service)
Users Python developers only Anyone - any language, any device
Setup Must install Python + all dependencies Just needs HTTP (curl, browser, any language)
Updates Every user must upgrade Deploy once, everyone gets the new version
Access control None API keys, rate limits, audit logs
Best for Reusable Python utilities Serving predictions to apps, dashboards, teams

Bottom line: Use a library when consumers are Python devs who want full control. Use an API when you want anyone to use your model without installing anything.

Learning Path

Lab Topic
Lab 00 REST fundamentals: HTTP verbs, status codes, URL design
Lab 01 Build your first API with FastAPI CRUD
Lab 02 Train a TF-IDF + Logistic Regression pipeline
Lab 03 Serve model predictions as a REST API
Lab 04 Test the API using pytest and FastAPI TestClient
Lab 05 Deploy with Render, HuggingFace Spaces, or Docker
Lab 06 Call HuggingFace, build a wrapper API, and evaluate LLM outputs

Total: ~3 hours

Quick Start

Option 1: GitHub Codespaces (Recommended)

Click Code → Codespaces → Create codespace on main. Everything is pre-configured via .devcontainer/devcontainer.json.

Option 2: Local Setup

# Clone the repo
git clone https://github.com/<your-username>/REST-API-Builder.git
cd REST-API-Builder

# Install uv (fast Python package manager) if you don't have it
pip install uv

# Install all dependencies (creates a .venv automatically)
uv sync --extra dev

# Start with Lab 00!

Folder Structure

REST-API-Builder/
├── README.md                    ← You are here
├── pyproject.toml               ← Dependencies, tool config (ruff, pytest)
├── render.yaml                  ← Render deployment config
├── Dockerfile                   ← Docker / HuggingFace Spaces config
├── .gitignore
├── .devcontainer/
│   └── devcontainer.json        ← Codespaces / Dev Container config
├── data/
│   └── clinical_notes.csv       ← 30 labelled clinical notes
├── models/
│   └── README.md                ← Generated model files go here
├── solutions/
│   ├── lab_00_answers.md        ← Quiz answers for Lab 00
│   ├── lab_01_challenge3.md     ← Solution for Lab 01 stretch challenge
│   ├── lab_03_challenges.md     ← Solutions for Lab 03 challenges
│   ├── lab_04_challenges.md     ← Solutions for Lab 04 challenges
│   ├── lab_05_challenges.md     ← Solutions for Lab 05 challenges
│   └── lab_06_challenges.md     ← Solutions for Lab 06 challenges
├── lab_00_rest_fundamentals/
│   └── README.md                ← REST theory + quiz
├── lab_01_your_first_api/
│   ├── app.py                   ← FastAPI CRUD server
│   └── README.md
├── lab_02_train_model/
│   ├── train.py                 ← Train the urgency classifier
│   └── README.md
├── lab_03_expose_model/
│   ├── app.py                   ← ML model as REST API
│   └── README.md
├── lab_04_test_api/
│   ├── test_api.py              ← pytest tests
│   └── README.md
├── lab_05_deploy/
│   ├── app.py                   ← Production-ready app
│   └── README.md
└── lab_06_llm_api/
    ├── app.py                   ← LLM wrapper API
    ├── test_app.py              ← API tests + LLM output evals
    └── README.md

Reference Materials

These resources are linked throughout the labs:

Resource Link
REST API Cheatsheet ByteByteGo
API Learning Roadmap ByteByteGo
What is an API IBM
API Lifecycle IBM
OpenAPI Intro Swagger
OpenAPI Specification Swagger
FastAPI Docs fastapi.tiangolo.com
Build FastAPI in Minutes Real Python
Different Types of APIs Postman Blog
HuggingFace Inference API huggingface.co/docs/api-inference
Training Script Inspiration ci-cd-template

License

This project is licensed under the MIT License.

Citation

If you use this template for teaching, research, or derivative work, please cite it:

@software{ojha_rest_api_builder,
  author  = {Ojha, Priyanka},
  orcid   = {https://orcid.org/0000-0002-6844-6493},
  title   = {REST-API-Builder: Hands-on REST API Tutorial for ML},
  url     = {https://github.com/po-DRA/API},
  version = {1.0.0},
  year    = {2026},
  license = {MIT}
}

Acknowledgements

This teaching material, code, documentation, diagrams, and README were developed by a human educator with AI assistance from Claude (Anthropic) , used as a pair-programming partner for debugging, drafting, and refining content throughout the project. It was tested by a real human.