🎬 Netflix-Style Recommendation System Demo

🧠 System Architecture

This project showcases a real-time, containerized recommendation pipeline using Docker Compose.
It integrates services like Spark, Flask API, Redis, Prometheus, and Grafana to simulate a Netflix-style recommendation system.

---

✅ Step 1: Show the System is Running

🎤 Narration:
“I’ve containerized and automated the entire Netflix-style recommendation pipeline using Docker Compose. Everything you see here is real-time and monitored.”

🔧 Command:

docker ps

🔍 Expected Output:

CONTAINER ID   IMAGE            COMMAND                  STATUS          PORTS
abc123         spark            "/bin/bash"              Up 5 minutes    ...
def456         jupyter          "start-notebook.sh"      Up 5 minutes    ...
ghi789         flask-api        "python app.py"          Up 5 minutes    5000/tcp
jkl012         redis            "docker-entrypoint.sh"   Up 5 minutes    6379/tcp
mno345         prometheus       "/bin/prometheus"        Up 5 minutes    9090/tcp
pqr678         grafana          "/run.sh"                Up 5 minutes    3000/tcp

🖼️ Screenshot Placeholder: Insert a screenshot of the terminal displaying the docker ps output.

---

✅ Step 2: Simulate a User Watching a Movie

🎤 Narration:
“Let’s say user U999 watches C1010, which is, say, Stranger Things. The system automatically predicts and caches the next likely show.”

🔧 Command:

curl -X POST http://localhost:5000/watched \
  -H "Content-Type: application/json" \
  -d '{"user_id": "U999", "content_id": "C1010"}'

🔍 Expected Response:

{
  "user_id": "U999",
  "watched": "C1010",
  "predicted_next": "C1045"
}

🖼️ Screenshot Placeholder: Insert a screenshot of the terminal displaying the curl response.

---

✅ Step 3: Retrieve Cached Recommendation

🎤 Narration:
“Now, the user comes back to the app and asks: what should I watch next? We instantly return the prediction from Redis.”

🔧 Command:

curl http://localhost:5000/recommend/U999

🔍 Expected Response:

{
  "user_id": "U999",
  "recommendation": "C1045",
  "cache_hit": true
}

🖼️ Screenshot Placeholder: Insert a screenshot of the terminal displaying the curl response.

---

✅ Step 4: Handle New User Scenario

🎤 Narration:
“If the user is new and we don’t have predictions yet, the system handles it gracefully with a fallback.”

🔧 Command:

curl http://localhost:5000/recommend/U1234

🔍 Expected Response:

{
  "user_id": "U1234",
  "recommendation": "C000",
  "cache_hit": false,
  "fallback_used": true
}

🖼️ Screenshot Placeholder: Insert a screenshot of the terminal displaying the curl response.

---

✅ Step 5: Monitor System Metrics with Prometheus

🎤 Narration:
“We also expose Prometheus metrics for system health and activity.”

🔧 Access:
Navigate to http://localhost:9090

🔍 Metrics to Observe:

• cache_hits
• cache_misses
• fallbacks_used

🖼️ Screenshot Placeholder: Insert a screenshot of the Prometheus dashboard displaying the relevant metrics.

---

✅ Step 6: Visualize Data with Grafana

🎤 Narration:
“Here’s the Grafana dashboard tracking live recommendations, cache efficiency, fallback trends, and more.”

🔧 Access:
Navigate to http://localhost:3000

🔐 Login Credentials:

• Username: admin
• Password: admin

*

---

🎁 Bonus Features

• Develop a simple HTML/JS frontend or CLI to simulate the watch and recommend flow.
• Display Spark job logs within JupyterLab or via Docker logs.
• Highlight the pipeline's extensibility for federated learning, edge caching, or A/B testing.