Data Pipeline & ETL Expert

English | Bahasa Indonesia

---

<a name="english"></a>

English

Orchestration & Integration

• Upstream Languages & Scripts: python-programming-expert, go-programming-expert, rust-programming-expert
• Transactional & Analytics DBs: database-orm-expert, search-engine-expert, vector-db-rag-expert
• Telemetry & Observability: data-telemetry-expert, logging-error-tracking-expert
• Scheduling & Queues: cron-scheduler-expert, async-queue-temporal-expert, event-driven-architect
• Lifecycle & Governance: zero-to-prod-orchestrator, session-memory-manager

Description

A specialized skill for building robust data architectures, Analytics Engineering, and ETL (Extract, Transform, Load) or ELT pipelines. It covers modern data stack orchestration (Airflow, Dagster), transformation tools (dbt), and high-performance OLAP databases (BigQuery, Snowflake, ClickHouse, DuckDB).

Trigger Conditions

• When designing reporting dashboards or analytics infrastructure for a SaaS.
• When moving large volumes of data from transactional databases (PostgreSQL/MySQL) to a data warehouse.
• When the user asks about "dbt", "Airflow", "ELT", or "Analytics Engineering".
• When building local or edge analytics using DuckDB.

Core Architectural Guidelines

1. ELT over ETL

Prefer Extract-Load-Transform (ELT) over traditional ETL.

• Extract & Load: Use tools like Airbyte or Fivetran to dump raw data directly into the Data Warehouse.
• Transform: Perform transformations inside the Data Warehouse using SQL (via dbt) to leverage the warehouse's massive compute power.

2. Analytics Engineering with dbt

Treat SQL like software engineering.

• Use dbt (Data Build Tool) to version control your SQL transformations.
• Implement tests (not_null, unique) on critical tables.
• Use Jinja templating in dbt to DRY up complex SQL queries.

3. Data Orchestration (Airflow vs Dagster)

• Apache Airflow: The industry standard for scheduling and monitoring complex DAGs (Directed Acyclic Graphs). Best for Python-heavy teams.
• Dagster: A modern alternative focused on data assets rather than just tasks. Use Dagster when you want better local testing and asset-driven lineage.

4. OLAP Database Selection

• BigQuery / Snowflake: Best for massive scale, fully managed cloud data warehousing.
• ClickHouse: Best for real-time, sub-second analytical queries on massive event streams.
• DuckDB: Best for local analytics, embedded analytical pipelines, or processing parquets in edge environments (Node.js/Python).

Implementation Examples

1. DuckDB In-Memory OLAP Pipeline (Node.js)

import duckdb from 'duckdb';

export async function processParquetAnalytics(parquetPath: string): Promise<Record<string, unknown>[]> {
  const db = new duckdb.Database(':memory:');
  const con = db.connect();

  return new Promise((resolve, reject) => {
    con.all(
      `SELECT
         date_trunc('day', timestamp) AS active_day,
         event_type,
         count(*) AS total_events,
         count(DISTINCT user_id) AS unique_users
       FROM read_parquet('${parquetPath}')
       GROUP BY 1, 2
       ORDER BY 1 DESC
       LIMIT 100`,
      (err, rows) => {
        if (err) return reject(err);
        resolve(rows);
      }
    );
  });
}

2. dbt Incremental Transformation Model (models/marts/fct_daily_usage.sql)

{{
  config(
    materialized = 'incremental',
    unique_key = ['tenant_id', 'metric_date']
  )
}}

SELECT
  tenant_id,
  CAST(event_time AS DATE) AS metric_date,
  count(event_id) AS total_api_calls,
  sum(tokens_consumed) AS total_tokens,
  current_timestamp() AS updated_at
FROM {{ ref('stg_telemetry_events') }}

{% if is_incremental() %}
  WHERE event_time >= (SELECT max(metric_date) FROM {{ this }})
{% endif %}

GROUP BY 1, 2

---

<a name="bahasa-indonesia"></a>

Bahasa Indonesia

Integrasi Orkestrasi

• Bahasa & Skrip Hulu: python-programming-expert, go-programming-expert, rust-programming-expert
• Database Transaksional & Analitik: database-orm-expert, search-engine-expert, vector-db-rag-expert
• Telemetri & Observabilitas: data-telemetry-expert, logging-error-tracking-expert
• Penjadwalan & Antrean: cron-scheduler-expert, async-queue-temporal-expert, event-driven-architect
• Siklus Hidup & Tata Kelola: zero-to-prod-orchestrator, session-memory-manager

Deskripsi

Panduan khusus untuk membangun arsitektur data yang kuat, Analytics Engineering, dan pipeline ETL/ELT. Mencakup orkestrasi (Airflow, Dagster), alat transformasi (dbt), dan database OLAP berkinerja tinggi (BigQuery, Snowflake, ClickHouse, DuckDB).

Kondisi Pemicu

• Saat merancang infrastruktur analitik atau dashboard pelaporan untuk SaaS.
• Saat memindahkan data bervolume besar dari database transaksional ke Data Warehouse.
• Saat membangun analitik lokal yang cepat menggunakan DuckDB.

Panduan Arsitektur Inti

1. ELT lebih disukai daripada ETL

• Extract & Load: Pindahkan data mentah (raw data) langsung ke Data Warehouse (menggunakan Airbyte/Fivetran).
• Transform: Lakukan transformasi data di dalam Data Warehouse menggunakan SQL (dbt) untuk memanfaatkan kekuatan komputasi gudang data yang masif.

2. Analytics Engineering dengan dbt (Data Build Tool)

Perlakukan transformasi data (SQL) layaknya rekayasa perangkat lunak. Gunakan dbt untuk version control, pengujian otomatis (not_null, unique), dan dokumentasi skema data Anda.

3. Orkestrasi Data (DAG)

Gunakan Apache Airflow atau Dagster untuk menjadwalkan dan memonitor alur kerja data yang kompleks (DAG). Dagster sangat direkomendasikan untuk pendekatan modern yang berpusat pada aset data (asset-driven orchestration).

4. Pemilihan Database OLAP

• BigQuery / Snowflake: Gudang data cloud fully-managed untuk analitik skala masif.
• ClickHouse: Sangat cepat untuk kueri analitik real-time. Cocok untuk data event stream/log.
• DuckDB: SQLite untuk analitik. Sangat cepat untuk memproses file Parquet atau CSV secara lokal maupun di lingkungan edge/serverless (via Python/Node.js).