Can I vibe code Amplitude?

Build a high-performance event tracking and behavioral analytics service designed for e-commerce stores.

The architecture consists of three core components:
1. Ingestion Ingest API & SDK: Node.js/TypeScript endpoint taking JSON payloads containing device_id, user_id, event_type, timestamp, session_id, and event_properties (JSONB). Ingest must run asynchronously, push to a Kafka or Redis Stream buffer, and acknowledge receipt within 20ms. Include an HTML/JS tracking snippet that batches micro-events every 2 seconds or on page unload.

2. Storage Layer: Set up a ClickHouse database with an `events` table using the MergeTree engine, partitioned by `toYYYYMM(timestamp)` and ordered by `(event_type, user_id, timestamp)`. Implement a consumer process to bulk-insert buffered events into ClickHouse every 1,000 records or 5 seconds.

3. Analytics Query API & Dashboard: Build a Next.js control panel that executes optimized ClickHouse SQL queries to calculate:
- Multi-step Funnel Analysis: Given an array of event names (e.g., ['view_item', 'add_to_cart', 'begin_checkout', 'purchase']), return step-by-step conversion rates and median time-to-convert within a configurable time window (e.g., 7 days).
- Behavioral Retention Matrix: N-day retention chart computing the percentage of users who performed event A (e.g., 'purchase') and returned to perform event B (e.g., 'view_item') on day N (0 to 30 days).
- User Activity Timeline: Query all chronological events for a specific user_id or device_id.

Out of scope: Complex A/B testing attribution, machine learning predictive cohorts, identity resolution across multi-domain cross-stitching, and native mobile SDK development. Focus strictly on web events, robust ClickHouse ingestion, and immediate funnel/retention rendering.

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