Can I vibe code Segmentify?

Build a lightweight ecommerce personalized search and recommendation API in Node.js (TypeScript) and Express, integrated with Meilisearch and PostgreSQL.

1. DATA MODEL & INGESTION:
- Create a Product schema: id, title, description, category, price, in_stock, tags, vector_embedding (array of floats).
- Create an OrderItem schema: order_id, product_id, created_at.
- Create a UserEvent schema: session_id, user_id, product_id, event_type ('view', 'cart_add', 'buy'), timestamp.
- Sync products to Meilisearch on creation or update.

2. SEARCH API:
- Endpoint: GET /api/search?q=query&session_id=string
- Query Meilisearch for keyword matches on title/description.
- Fetch user's last 5 viewed products from Redis using session_id. If recent views exist, boost matching categories in the search results.
- Return top 20 ranked product JSON objects under 50ms.

3. RECOMMENDATION ENGINE:
- Endpoint: GET /api/recommendations/frequently-bought-together?product_id=string
- Query PostgreSQL order history to return top 4 products frequently ordered in the same transaction as product_id.
- Endpoint: GET /api/recommendations/similar?product_id=string
- Perform vector cosine similarity search against product_embedding in pgvector or Meilisearch to return top 6 visual/semantic matches.

4. EVENT TRACKING & MERCHANDISING OVERRIDES:
- Endpoint: POST /api/events to ingest clickstream data and append product_ids to a Redis list keyed by session_id (expire after 24h).
- Allow manual overrides via a Merchandising Table (pinned_products, boosted_brands) that inject specified products into top search ranks.

5. OUT OF SCOPE:
- Complex email/push notification queue processing.
- Frontend UI widgets (provide clean JSON responses only).

$ each button prefixes agent-specific run instructions · build your own product, never copy proprietary code, trademarks or designs