Can I vibe code Synerise?

Build a lightweight Customer Data Platform and Recommendation Engine MVP using Next.js, Node.js, PostgreSQL with Pgvector, and ClickHouse. 

1. DATA MODEL & INGESTION:
- Create a JavaScript browser snippet that tracks events: page_view, product_view, add_to_cart, purchase.
- Build a high-throughput API endpoint `/api/v1/events` using Node.js that accepts JSON event payloads and asynchronously writes them to a ClickHouse `events` table with schema: event_id, user_id, session_id, event_type, product_id, timestamp, metadata (JSON).
- Maintain a PostgreSQL `profiles` table aggregating user_id, email, lifetime_value, last_seen, and category_affinities.

2. AI RECOMMENDATIONS & VECTOR SEARCH:
- Create a cron job that generates product embeddings using OpenAI text-embedding-3-small and stores them in PostgreSQL using Pgvector.
- Build a recommendation endpoint `/api/v1/recommendations` that takes a `user_id` or `product_id` and performs cosine similarity queries to return top 6 recommended products based on co-occurrence in cart/views and vector distance.

3. EVENT AUTOMATION QUEUE:
- Set up BullMQ with Redis to process real-time events.
- If an `add_to_cart` event is logged without a corresponding `purchase` event within 60 minutes, trigger a webhook to SendGrid/Klaviyo with the abandoned cart item payload.

4. ADMIN DASHBOARD:
- Build a React dashboard showing real-time ingested events/sec, active user profiles, top recommended items, and conversion rate of recommendation clicks.

5. OUT OF SCOPE:
- Do not build dynamic drag-and-drop visual workflow builders.
- Do not handle heavy real-time neural search model re-training; rely on static daily embedding updates.

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