Data Science — Model Training

Deployment options

1. REST API: Flask/FastAPI
2. Serverless: Lambda/Cloud Functions
3. Containers: Docker/Kubernetes
4. ML platforms: SageMaker/AI Platform

Flask deployment

from flask import Flask, request, jsonify
import pickle
import numpy as np

app = Flask(__name__)
model = pickle.load(open("model.pkl", "rb"))

@app.route("/predict", methods=["POST"])
def predict():
    data = request.get_json()
    features = np.array(data["features"]).reshape(1, -1)
    prediction = model.predict(features)[0]
    return jsonify({"prediction": prediction.tolist()})

if __name__ == "__main__":
    app.run(host="0.0.0.0", port=5000)

FastAPI deployment

from fastapi import FastAPI
from pydantic import BaseModel
import pickle

app = FastAPI()
model = pickle.load(open("model.pkl", "rb"))

class PredictionRequest(BaseModel):
    features: list

@app.post("/predict")
def predict(request: PredictionRequest):
    prediction = model.predict([request.features])
    return {"prediction": prediction.tolist()}

Docker

FROM python:3.9-slim
WORKDIR /app
COPY requirements.txt .
RUN pip install -r requirements.txt
COPY . .
CMD ["python", "app.py"]

Best practices

1. Use versioning
2. Implement monitoring
3. Add health checks
4. Load testing

Mini Practice

1. Deploy with Flask
2. Create FastAPI endpoint
3. Containerize with Docker
4. Test deployment

Up Next

Continue with AutoML - Automated machine learning.