💘 Crushie

AI Dating Coach for Students

Learn communication skills through AI-powered profile analysis, conversation coaching, and real-time feedback.

---

Overview

Crushie is an AI-powered communication coach that helps students build real social skills through profile analysis and personalized feedback—like Duolingo, but for dating.

Upload a dating profile screenshot → AI analyzes their communication style → Get personalized conversation starters and date ideas.

---

Key Features

Profile Analyzer - Upload a screenshot, get instant insights:

• Communication style prediction (playful, intellectual, direct, adventurous, shy)
• 8 personalized conversation starters
• 3 date suggestions with compatibility scoring
• Privacy-first: only hashes stored, never raw images

Educational Framework - Learn social skills through practice:

• Track your communication progress
• Build confidence with AI feedback
• Practice in a judgment-free space

Dual AI System - Enterprise-grade reliability:

• Primary: Google Gemini 2.0 Flash (fast, cost-efficient)
• Fallback: Microsoft Azure OpenAI Phi-4 (98.5% uptime)

---

Tech Stack

Frontend: Next.js 16, TypeScript, tRPC, Framer Motion
Backend: Node.js, Express, PostgreSQL (Supabase), Drizzle ORM
AI: Google Gemini 2.0 Flash + Azure OpenAI Phi-4
Auth: Clerk with JWT-based Row Level Security
Infrastructure: Docker, Redis caching

---

Quick Start

Prerequisites

• Node.js 22+
• Docker Desktop
• Supabase CLI

Installation

# Install dependencies
npm install

# Start Supabase
npx supabase start

# Start LLM service
cd apps/llm
docker compose up -d

# Start web client
cd apps/web-client
npm run dev

Visit http://localhost:3000/analyze-profile

---

Environment Variables

Web Client (apps/web-client/.env.local)

# Clerk Authentication
NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY=your_key
CLERK_SECRET_KEY=your_secret

# Database
DATABASE_URL=postgresql://...
DIRECT_URL=postgresql://...

# LLM Service
LLM_URL=http://localhost:3001

LLM Service (apps/llm/.env)

# AI Providers
GEMINI_API_KEY=your_key
AZURE_OPENAI_ENDPOINT=https://your-resource.cognitiveservices.azure.com/
AZURE_OPENAI_KEY=your_key
AZURE_OPENAI_DEPLOYMENT=Phi-4-mini-instruct
AZURE_OPENAI_API_VERSION=2024-05-01-preview

# Caching
REDIS_URL=redis://localhost:6379

---

Project Structure

crushie/
├── apps/
│   ├── web-client/          # Next.js web app
│   │   ├── src/app/         # App router pages
│   │   └── src/services/    # tRPC procedures
│   └── llm/                 # AI microservice
│       ├── src/routes/      # API endpoints
│       └── src/lib/         # Prompt templates
└── supabase/
    └── migrations/          # Database schema

---

Key Scripts

# Development
npm run dev:web              # Start web client
npm run dev:llm              # Start LLM service

# Database
npx supabase start           # Start local database
npx supabase db reset        # Reset database with migrations

# Docker
cd apps/llm && docker compose up -d    # Start LLM service

---

How It Works

1. Upload - User uploads dating profile screenshot
2. Hash - Client-side SHA-256 hashing (privacy-first)
3. Analyze - AI analyzes communication style via Gemini Vision API
4. Generate - LLM creates conversation starters and date suggestions
5. Display - Animated results with copy-to-clipboard features

---

Built With

• PatriotAI - Prompt design and concept development
• Google Gemini 2.0 Flash - Multimodal vision analysis
• Microsoft Azure OpenAI - Enterprise reliability fallback
• Supabase - PostgreSQL with Row Level Security
• Docker - Containerized LLM service

---

License

This project is proprietary. All rights reserved.

---

Team

• Bao Tran https://github.com/BaoT1301
• Lam Anh https://github.com/anhlamtruong
• Mai Tran https://github.com/tranthanhmai2006
• Nguyen Ho https://github.com/hodangkhoinguyen

---

Built at PatriotHacks 2026 🚀