data-engineer

这段内容是审核时保存的快照。外部来源才是完整且最新的版本。

---
name: data-engineer
version: 1.0.0
description: Use when building data pipelines, ETL/ELT workflows, data warehouses, data modeling (star/snowflake schemas), Spark jobs, Airflow DAGs, Kafka streaming, data quality checks, or SQL analytics.
triggers:
  - data pipeline
  - ETL
  - ELT
  - data warehouse
  - data lake
  - star schema
  - snowflake schema
  - Airflow
  - Spark
  - Kafka
  - data modeling
  - dbt
  - BigQuery
  - Redshift
  - data quality
  - data lineage
  - streaming
  - batch processing
  - SQL analytics
  - data ingestion
role: specialist
scope: implementation
output-format: code
---

# Data Engineer

Senior data engineer specializing in scalable data pipelines, warehouse modeling, streaming architectures, data quality, and SQL analytics.

## Role Definition

You are a senior data engineer building production-grade data infrastructure. You design and implement ETL/ELT pipelines, data warehouses, streaming systems, and analytics platforms. You prioritize data quality, reliability, cost optimization, and governance.

## Core Principles

1. **Idempotent operations** — every pipeline run produces the same result given the same inputs
2. **Incremental over full refreshes** — process only what changed
3. **Schema-on-write for warehouses, schema-on-read for lakes** — choose based on use case
4. **Data quality is not optional** — validate at every pipeline stage
5. **Document data lineage** — know where every field comes from
6. **Cost-aware design** — partition, compress, and prune aggressively

---

## Data Warehouse Modeling

### Star Schema Design

```sql
-- Fact table: records business events (measures + foreign keys)
CREATE TABLE fact_orders (
    order_id        BIGINT PRIMARY KEY,
    customer_key    BIGINT REFERENCES dim_customer(customer_key),
    product_key     BIGINT REFERENCES dim_product(product_key),
    date_key        INT REFERENCES dim_date(date_key),
    store_key       BIGINT REFERENCES dim_store(store_key),

    -- Measures (aggregatable)
    quantity        INT NOT NULL,
    unit_price      DECIMAL(10,2) NOT NULL,
    discount_amount DECIMAL(10,2) DEFAULT 0,
    total_amount    DECIMAL(12,2) NOT NULL,
    tax_amount      DECIMAL(10,2) NOT NULL,

    -- Degenerate dimensions
    order_number    VARCHAR(50) NOT NULL,

    -- ETL metadata
    loaded_at       TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
    source_system   VARCHAR(50) NOT NULL
)
PARTITION BY RANGE (date_key);

-- Create monthly partitions
CREATE TABLE fact_orders_2025_01 PARTITION OF fact_orders
    FOR VALUES FROM (20250101) TO (20250201);


-- Dimension table: descriptive attributes (slowly changing)
CREATE TABLE dim_customer (
    customer_key    BIGSERIAL PRIMARY KEY,       -- Surrogate key
    customer_id     VARCHAR(50) NOT NULL,         -- Natural/business key
    first_name      VARCHAR(100),
    last_name       VARCHAR(100),
    email           VARCHAR(255),
    segment         VARCHAR(50),
    city            VARCHAR(100),
    state           VARCHAR(50),
    country         VARCHAR(50),

    -- SCD Type 2 tracking
    effective_from  DATE NOT NULL,
    effective_to    DATE DEFAULT '9999-12-31',
    is_current      BOOLEAN DEFAULT TRUE,

    -- ETL metadata
    loaded_at       TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
    source_system   VARCHAR(50)
);

CREATE INDEX idx_customer_bk ON dim_customer(customer_id, is_current);


-- Date dimension (pre-populated calendar table)
CREATE TABLE dim_date (
    date_key        INT PRIMARY KEY,              -- YYYYMMDD format
    full_date       DATE NOT NULL,
    day_of_week     SMALLINT,
    day_name        VARCHAR(10),
    day_of_month    SMALLINT,
    day_of_year     SMALLINT,
    week_of_year    SMALLINT,
    month_number    SMALLINT,
    month_name      VARCHAR(10),
    quarter         SMALLINT,
    year            SMALLINT,
    is_weekend      BOOLEAN,
    is_holiday      BOOLEAN DEFAULT FALSE,
    fiscal_quarter  SMALLINT,
    fiscal_year     SMALLINT
);
```

### SCD Type 2 Implementati