A Tinder-like fashion discovery app that learns your style preferences through swiping and provides personalized clothing recommendations using AI.

Imagine Tinder for Fashion - but smarter! Users swipe through clothing items to express their preferences, and the AI learns their style to recommend products they'll actually love.

• Discovery Mode (Swipes 1-20): Browse 6,000+ fashion items from 8+ premium brands
• AI Mode (Swipe 21+): Get personalized recommendations powered by SuperMemory AI
• Real-time Learning: Every swipe is instantly saved and analyzed

• Frontend: Streamlit (Python web app)
• AI/ML: SuperMemory API for preference learning and recommendations
• Data: 6,000+ products with AI-generated clothing descriptions (LLaVA Vision Model)
• Brands: Alo Yoga, Gymshark, NAKD, Princess Polly, Vuori, and more

Slapp/
├── src/                                 # Reusable libraries/modules
│   └── supermemory/
│       ├── __init__.py
│       └── client.py                    # Shared SuperMemory client (env-driven)
├── scripts/
│   └── ingestion/
│       ├── supermemory_batch_push.py    # Bulk upload products to SuperMemory
│       ├── supermemory_push_async.py    # Async single upload example
│       ├── supermemory_helper.py        # Simple single upload example
│       └── supermemory_search.py        # Search testing helper
├── pipelines/
│   └── vision/
│       └── ViT_Img_Descriptor.py        # Vision descriptor pipeline
├── streamlit-product-display/           # Main Streamlit application
│   ├── src/
│   │   ├── app.py                       # Core app with swiping logic & UI
│   │   ├── user_memory.py               # Save preferences to SuperMemory
│   │   ├── get_user_preference.py       # Retrieve user preferences
│   │   ├── query_main_memory.py         # AI recommendation engine
│   │   ├── components/
│   │   │   └── product_card.py          # Product display components
│       └── utils/
│         └── data_loader.py           # CSV data loading utilities
│   
├── utils/
│   ├── preprocess.py                    # Unify brand-specific CSV files
│   └── download_images.py               # Download images helper
├── data/                                # Brand-specific product datasets
│   ├── alo_yoga_products.csv
│   ├── gymshark_products.csv
│   └── ...
├── archive_data/
│   └── ...
├── final_products_complete.csv          # Main product dataset (6k items)
├── requirements.txt                     # Root deps for scripts/libs
├── .gitignore
└── README.md

1. Download and preprocess images and data

• Run your image download and preprocessing utilities to build input CSVs:
  • utils/download_images.py
  • utils/preprocess.py

2. Generate clothing descriptions (Vision pipeline)

• Use the ViT descriptor to generate clothing_features and produce the consolidated dataset.
• Output: final_products_complete.csv at project root.

python pipelines/vision/ViT_Img_Descriptor.py

3. Ingest to SuperMemory (Batch)

• Ensure .env has your API key (see Setup below)
• Install root deps and run the batch ingestion script:

pip install -r requirements.txt
python scripts/ingestion/supermemory_batch_push.py

4. Run the Streamlit App

• Install app deps and launch:

streamlit run app.py

1. Install dependencies

   cd streamlit-product-display
   pip install -r requirements.txt

2. Configure API key

• Streamlit app (preferred): create .streamlit/secrets.toml with:

SUPERMEMORY_API_KEY = "your_supermemory_api_key_here"

• CLI/ingestion scripts: set an environment variable or use a project .env file at Slapp/.env with:

SUPERMEMORY_API_KEY=your_supermemory_api_key_here

3. Run the app (from project app root)

   cd streamlit-product-display
   streamlit run src/app.py

4. Run ingestion scripts

Ensure SUPERMEMORY_API_KEY is available in your environment (via Slapp/.env or shell export), then run scripts from anywhere, e.g.:

python scripts/ingestion/supermemory_batch_push.py

3. Start discovering fashion
   • Open browser to http://localhost:8501
   • Swipe ❤️ Like, ⭐ Super Like, or 👎 Pass on clothing items
   • Watch the AI learn your style and provide personalized recommendations

1. Browse & Learn: Users swipe through curated fashion items
2. Preference Capture: Each swipe is saved to SuperMemory with detailed product metadata
3. AI Analysis: SuperMemory analyzes collective preferences to understand user style
4. Smart Recommendations: AI queries the product database to find matching items
5. Continuous Improvement: More swipes = better recommendations

• API keys are read from Streamlit st.secrets. Configure via .streamlit/secrets.toml. The shared client is in src/supermemory/client.py.
• CLI scripts under scripts/ingestion/ locate CSVs relative to the project root, so you can run them from any directory.
• When running Streamlit, app.py prepends the project src/ to sys.path so shared code is importable.

• Session-based: Each user gets a private style profile
• Background AI Building: Recommendations prepared while you browse
• Fallback Systems: App never breaks - switches to random products if needed
• Rich Product Data: AI-generated clothing descriptions enable semantic matching