ML API - AI-Powered Text Generation API

A production-ready FastAPI application that provides AI text generation with response caching and comprehensive monitoring.

Features

• 🤖 AI Text Generation - Uses Google's Flan-T5-Small model
• ⚡ Response Caching - 1000x faster for repeat queries
• 📊 Metrics Tracking - Real-time performance monitoring
• 🔍 Comprehensive Logging - Detailed request/response tracking
• ✅ Input Validation - Prevents malformed requests
• 💚 Health Checks - Production monitoring ready

Tech Stack

• FastAPI - Modern Python web framework
• Transformers (Hugging Face) - AI model integration
• Flan-T5-Small - 80M parameter text generation model
• Python 3.12

Quick Start

Prerequisites

• Python 3.8+
• 8GB+ RAM (for model)

Installation

1. Clone the repository:

git clone <your-repo-url>
cd ml-api-project

2. Create virtual environment:

python3 -m venv venv
source venv/bin/activate  # On Mac/Linux

3. Install dependencies:

pip install fastapi uvicorn transformers torch

4. Run the API:

uvicorn main:app --reload

5. Open your browser:

http://127.0.0.1:8000/docs

API Endpoints

GET /

Root endpoint - confirms API is running

GET /health

Health check endpoint for monitoring

GET /metrics

Returns API usage statistics:

• Total requests
• Cache hit rate
• Average inference time
• And more...

POST /generate

Main text generation endpoint

Request Body:

{
  "prompt": "What is machine learning?",
  "max_length": 100
}

Response:

{
  "prompt": "What is machine learning?",
  "response": "Machine learning is a method of data analysis...",
  "model": "flan-t5-small",
  "inference_time_seconds": 5.59,
  "cached": false
}

Performance

• First request: ~5-10 seconds (runs AI model)
• Cached requests: ~0.01 seconds (instant)
• Cache hit rate: Typically 60%+ in production

Example Usage

Using curl:

curl -X POST "http://127.0.0.1:8000/generate" \
  -H "Content-Type: application/json" \
  -d '{"prompt": "What is Python?", "max_length": 100}'

Using Python:

import requests

response = requests.post(
    "http://127.0.0.1:8000/generate",
    json={"prompt": "What is Python?", "max_length": 100}
)

print(response.json())

Logging

The API uses comprehensive logging with emoji markers:

• 🔵 New request received
• 📝 Prompt details
• 💾 Cache hit/miss
• 🤖 AI model running
• ✅ Response generated
• ⚠️ Warnings/errors

Future Improvements

• Deploy to AWS ECS/Lambda
• Add Redis for persistent caching
• Implement rate limiting
• Add authentication
• Support for multiple models
• Streaming responses
• OpenAPI/Swagger customization

Project Structure

ml-api-project/
├── main.py              # Main application code
├── README.md            # This file
├── requirements.txt     # Python dependencies
└── venv/               # Virtual environment

Author

Built as part of an AI engineering portfolio project.

License

MIT License