Build a Data Warehouse: Kafka → PostgreSQL ELT Pipeline w/ dbt + Superset (Multi-Tenant SaaS)

---

Project Overview: We are a SaaS company with a multi-tenant architecture and Kafka-based event pipelines. We are looking for a skilled data engineer to implement an ELT system using: - Kafka (event source) - Kafka Connect (for streaming data into a central warehouse) - PostgreSQL (as the reporting warehouse) - dbt (for transformation and normalization) - Superset (for reports and non-developer-friendly reporting) Project Goals: - Deliver a reliable Kafka → PostgreSQL pipeline using Kafka Connect - Build clean analytics models using dbt - Enable our non-technical staff to build reusable, tenant-aware reports in Superset Current Architecture: - Each of the 23 microservices (Account, Commerce, etc.) has its own PostgreSQL DB - Each tenant has its own schema/namespace - All services emit Kafka events when DB records change - We are deploying a dedicated PostgreSQL instance as a reporting warehouse Scope of Work Phase 0 – Warehouse Schema Design - Propose and define the initial schema structure for the central reporting PostgreSQL warehouse - Includes raw Kafka-ingested tables and normalized dbt models - Should support multi-tenant structure (e.g., tenant_id column, single schema vs schema-per-tenant) - Design for scalability across services (e.g., account_users, commerce_orders) - Set up required schemas/namespaces (e.g., raw, staging, analytics) - Ensure naming conventions and structure are dbt- and Superset-friendly - Review schema plan with our team before implementation - Document decisions and structure for handoff Phase 1 – Initial Backfill (One-Time Load) - Connect to each service PostgreSQL database (e.g., Account, Commerce) - Extract historical data from each tenant schema within those databases - Load the extracted data into the appropriate tables in the reporting PostgreSQL warehouse - Normalize field types and formats to match Kafka-ingested data for consistency - Ensure tenant_id and source_service fields are included - Automate this step as a repeatable script or process in case re-runs are needed - Document the process clearly Phase 2 – Kafka Connect to PostgreSQL - Set up Kafka Connect using open-source connectors - Configure sinks for multiple topics into the central PostgreSQL warehouse - Ensure tenant context (tenant_id) is preserved in target tables - Document topic-to-table mappings Phase 3 – dbt Modeling - Create and configure a clean, modular dbt project - Write initial models to transform raw event data into curated tables (E.g., users, orders, subscriptions) - Normalize fields across services where applicable - Add documentation and basic data tests Phase 4 – Superset Charts/Table Reports - Connect Superset to the reporting warehouse - Create example dashboards with filters for: - Date ranges - Tenant selection - Service-specific views - Recommend best practices for access control (row-level or embedding strategies) Desired Experience - Kafka Connect (sink connectors, JDBC, PostgreSQL) - dbt (including incremental models, Jinja templating, modular structure) - PostgreSQL (for schema design and indexing in a warehouse context) - Superset configuration and dashboard setup - Experience with multi-tenant data structures is highly preferred

Keyword: E-commerce Development

dbt PostgreSQL ETL Pipeline Apache Kafka Apache Superset