Curate
Curate: Where AI Meets Human Connection
Curate is a modern matchmaking platform that combines the personal touch of human matchmakers with the precision of AI-powered compatibility algorithms. Our mission is to help individuals find meaningful relationships through a carefully curated process.
The Challenge: Overhauling a Fundamentally Broken System
The Curate Matchmaking project quickly evolved from feature improvements to a complete system overhaul. The core problems were systemic, stemming from a poorly integrated matchmaking process.
Original System Flaws:
- Overcomplicated Matching Logic.
- Original Feature Extraction limitations.
- Architectural Problems – tightly coupled services.
- Legacy Database Usage – inefficient queries.
The Solution: A Full-Stack AI Revamp
Our approach was a full-stack AI revamp, tackling the entire system instead of applying reactive patches. Key changes included:
Key Improvements:
- Image Processing Enhancement (Facenet architecture).
- LLM-Powered Questionnaire Transformation.
- Dynamic Data Pipeline – continuous learning and adaptation.
Advanced Matchmaking System
- Multimodal Feature Extraction.
- Hybrid Search System.
- Enhanced Scoring System.
- AI-Powered Analysis with advanced language models.
Unlocking Deeper Connections: The Curate Advantage
Curate offers a unique blend of AI and human expertise to deliver exceptional matchmaking results.
Key Benefits:
- AI-Enhanced Human Matchmaking: The personal touch with sophisticated AI algorithms for optimal matching.
- Intelligent Match Filtering: advanced filtering systems considering preferences.
- Scalable Architecture: Built to handle growth with modern cloud infrastructure.
Technical Implementation: From Reactive to Proactive
Here's how we tackled the core issues and transformed the Curate matchmaking system:
Streamlined Matching Pipeline:
async def build_vector_search_pipeline(user: UserDocument):
vector_search_stage = {
"$vectorSearch": {
"index": "default",
"path": "facenet_embedding",
"queryVector": user.fit_finder_centroid,
"limit": 25,
"numCandidates": 5000
}
}
- More robust data integration
- Dynamic search capabilities.
External API's:
- Utilized SendInBlue, Twilio, Retell, Slack, OpenAI and Anthropic api's.
- The proper implementation to help speedup the matchmaking platform.
Database:
- MongoDB with Beanie ODM.
- We improved the queries so the matchmaking platform can search people more efficiently.
Key Features Deep Dive
A closer look at the core functionalities that drive Curate's success.
Intelligent Matching System:
- Preference learning from user interactions.
- Multi-factor compatibility calculation.
User Journey Management:
- Profile completion and verification.
- Introduction management and match presentation.
Future Enhancements: The Road Ahead
We are committed to continuously improving Curate. Our future plans include:
- Mobile app development for enhanced accessibility.
- Enhanced matching algorithms for even better compatibility.
Results: Significant Improvements Across the Board
The new system dramatically improved performance.
Key Metrics:
- Average match generation time: Improved from 3.2s to 0.8s.
- Hard filter accuracy: Improved from ~60% to 100%.
- User satisfaction rate: Improved from ~50% to ~75%.
Conclusion: Transforming Matchmaking with AI
By implementing these core principles, Curate Matchmaking has moved from a broken state towards a proactive system, and system performance improved.
Ready to Transform Your Platform?
Whether you're looking to build a new matchmaking system or enhance your existing platform, we bring the technical expertise and innovative approach to make it happen.