Free Tier Deployment Options

Deploy your Text Classification API on free tier platforms with optimized configurations.

Platform Comparison

Platform	Free Tier Limits	Build Time	Cold Start	Scaling
Render	750 hours/month	Fast	Medium	Manual
Railway	$5/month credit	Fast	Fast	Auto
Fly.io	3 shared CPUs	Medium	Fast	Manual
Heroku	550 hours/month	Slow	Slow	Auto

Render (Recommended)

Setup

1. Connect Repository

   # Push your code to GitHub
   git add .
   git commit -m "Initial commit"
   git push origin main
   

2. Create Render Service

3. Go to Render Dashboard
4. Click "New" → "Web Service"
5. Connect your GitHub repository
6. Configure build settings:

Build Settings

# render.yaml (optional)
services:
  - type: web
    name: text-classifier-api
    env: python
    buildCommand: "pip install -r api/api_requirements.txt"
    startCommand: "cd api && python main.py"
    envVars:
      - key: PORT
        value: 10000
      - key: MODEL_PATH
        value: final_best_model.pkl
      - key: VECTORIZER_PATH
        value: tfidf_vectorizer.pkl

Environment Variables

PORT=10000
MODEL_PATH=final_best_model.pkl
VECTORIZER_PATH=tfidf_vectorizer.pkl
MAX_BATCH_SIZE=25
ENABLE_METRICS=false

File Structure for Render

your-repo/
├── api/
│   ├── main.py
│   ├── api_requirements.txt
│   └── Dockerfile (optional)
├── final_best_model.pkl
├── tfidf_vectorizer.pkl
└── render.yaml

Railway

Setup

1. Install Railway CLI

   npm install -g @railway/cli
   railway login
   

2. Deploy

   cd api
   railway init
   railway up
   

3. Environment Variables

   railway variables set PORT=8000
   railway variables set MODEL_PATH=../final_best_model.pkl
   railway variables set VECTORIZER_PATH=../tfidf_vectorizer.pkl
   

Fly.io

Setup

1. Install Fly CLI

   # Download from https://fly.io/docs/getting-started/installing-flyctl/
   flyctl auth login
   

2. Create App

   cd api
   flyctl launch
   

3. Configure fly.toml

   app = "text-classifier-api"
   kill_signal = "SIGINT"
   kill_timeout = 5
   processes = []
   
   [env]
     PORT = "8080"
   
   [experimental]
     allowed_public_ports = []
     auto_rollback = true
   
   [[services]]
     http_checks = []
     internal_port = 8080
     processes = ["app"]
     protocol = "tcp"
     script_checks = []
   
     [services.concurrency]
       hard_limit = 25
       soft_limit = 20
       type = "connections"
   
     [[services.ports]]
       force_https = true
       handlers = ["http"]
       port = 80
   
     [[services.ports]]
       handlers = ["tls", "http"]
       port = 443
   
     [[services.tcp_checks]]
       grace_period = "1s"
       interval = "15s"
       restart_limit = 0
       timeout = "2s"
   

Heroku

Setup

1. Install Heroku CLI

   # Download from https://devcenter.heroku.com/articles/heroku-cli
   heroku login
   

2. Create App

   cd api
   heroku create your-app-name
   

3. Configure Environment

   heroku config:set MODEL_PATH=../final_best_model.pkl
   heroku config:set VECTORIZER_PATH=../tfidf_vectorizer.pkl
   

4. Deploy

   git push heroku main
   

Optimization Tips

Memory Optimization

1. Reduce Model Size

   # Use joblib compression
   joblib.dump(model, 'model.pkl', compress=3)
   

2. Limit Batch Size

   export MAX_BATCH_SIZE=25
   

3. Disable Metrics in Production

   export ENABLE_METRICS=false
   

Cold Start Optimization

1. Pre-load Models

2. Models are loaded on startup to reduce cold start time

3. Use Connection Pooling

4. Built-in async processing reduces connection overhead

5. Optimize Dependencies

6. Minimal dependencies for faster installs

Cost Optimization

1. Monitor Usage

   # Check memory usage
   curl https://your-api.com/health
   

2. Scale Down When Possible

3. Free tiers often auto-scale based on traffic

4. Use CDN for Static Assets

5. Not applicable for API, but consider for frontend

Troubleshooting

Common Issues

1. Memory Limit Exceeded

   Solution: Reduce MAX_BATCH_SIZE, disable metrics
   

2. Timeout Errors

   Solution: Increase timeout limits, optimize model loading
   

3. Model Loading Failures

   Solution: Ensure model files are in correct paths
   

Monitoring

• Check /health endpoint for service status
• Monitor response times and error rates
• Set up alerts for high memory usage

Migration Between Platforms

From Render to Railway

# Export environment variables
railway variables set $(heroku config --json | jq -r 'to_entries[] | "\(.key)=\(.value)"')

Backup Strategy

# Download model files before migration
curl -O https://your-api.com/final_best_model.pkl
curl -O https://your-api.com/tfidf_vectorizer.pkl