AI Microservices Lab

Vector Recommendations for Cuculi Social Dining

Explore how Cuculi pairs diners with unforgettable events using MongoDB vector similarity and AI-personalized outreach. Toggle between the user-to-event recommender and the event-to-user targeting engine, then inspect the service architecture.

AI Services Control Center

Switch between the overview, user recommender, and event targeting flows. Toggle the architecture map to see how the services snap together.

AI-Powered Recommendation Engine

Two cooperating microservices keep Cuculi's community calendar vibrant. Feed in a diner profile to curate events with custom outreach, or start with a marquee event and assemble the perfect guest list.

User Recommendation Service

Vector matching aligns diner personalities with open tables, while AI copywriters keep outreach human and inviting.

  • Cosine similarity across vectorized tastes and vibes
  • OpenAI-tailored RSVP prompts based on context
  • Compatibility scoring and reason tagging

Event Targeting Service

Reverse the lens: start with a headline event and summon the right mix of regulars, rising stars, and churn-risk diners.

  • Dynamic segmentation across engagement tiers
  • Retention-aware messaging for churn prevention
  • Contextual targeting reasons surfaced for operators

How the Loop Runs

1

Vector embeddings hydrate diner and event profiles.

2

MongoDB Atlas performs cosine similarity lookups.

3

OpenAI composes human-ready outreach copy.

4

Operators review ranked matches and launch campaigns.

API Contracts

POST /api/users/{user_id}/recommendations

{
  "limit": 10,
  "filters": {
    "neighborhood": "optional",
    "price_range": "optional",
    "cuisine_type": "optional"
  }
}

Response

{
  "recommendations": [
    {
      "event_id": "...",
      "compatibility_score": 85,
      "personalized_message": "...",
      "reasons": ["..."]
    }
  ]
}

Campaign Targeting

POST /api/events/{event_id}/targeting

{
  "limit": 20,
  "segments": ["High-Value Active", "New Users"],
  "min_compatibility": 70
}

Response

{
  "target_users": [
    {
      "user_id": "...",
      "compatibility_score": 92,
      "personalized_message": "...",
      "targeting_reasons": ["..."]
    }
  ]
}
Stack MongoDB Vector Search OpenAI GPT-4 Node.js React & Native Clients
Dive into the full architecture brief