Key Responsibilities

• Design data models, dataset layouts, and pipeline architectures for new data domains within the established platform architecture.
• Build and maintain ELT pipelines (Airbyte ingestion → dbt transformations → curated marts) orchestrated in Dagster.
• Operate Apache Iceberg tables in production: partitioning strategy, file compaction, snapshot expiration and retention, schema evolution, and time-travel-based reprocessing.
• Operate and extend CDC ingestion with Debezium and Kafka: connector configuration, PostgreSQL logical replication (WAL / replication slots), schema registry, idempotent sinks, backfill and replay procedures.
• Tune Trino performance: storage layout, table statistics, query plans, resource groups.
• Engineer data quality: dbt tests, data contracts, freshness / volume / schema-drift monitoring, duplicate detection, missing-value handling; maintain lineage and metadata in OpenMetadata.
• Ensure reproducibility and traceability through Iceberg snapshots and versioned dbt models.
• Build feature and embedding pipelines serving AI services (entity resolution, deduplication, vector stores).
• Implement governed pipelines: multi-tenant isolation, data classification, access-control integration (Keycloak / OPA), audit logging, and compliance with Vietnamese data-residency requirements.
• Contribute to architectural evolution: evaluate alternatives, write ADRs, participate in design reviews (e.g., streaming expansion, catalog or vector-store migration).
• Own production reliability: SLOs, monitoring, runbooks, incident triage, and root-cause analysis for data pipelines.
• Collaborate with Backend, AI, and Analytics teams to deliver reliable data products.

Requirements

• Bachelor's degree in Computer Science, Information Technology, or a related field — or equivalent practical experience.
• 5+ years as a Data Engineer with end-to-end production ownership.
• Strong SQL and Python.
• Hands-on experience designing Data Lake / Lakehouse and Data Warehouse architectures and data models — and the judgment to work effectively within an established architecture.
• Proven ELT/ETL pipelines across heterogeneous sources (databases, APIs, files, event streams).
• Production experience with at least one open table format — Apache Iceberg strongly preferred; Delta Lake or Hudi acceptable with commitment to transition.
• Hands-on experience with a distributed SQL engine (Trino / Presto or comparable) and dbt or an equivalent transformation framework.
• Production orchestration experience — Dagster preferred; Airflow / Prefect acceptable with commitment to transition.
• CDC experience (Debezium or equivalent) and working knowledge of PostgreSQL logical replication.
• Experience building feature pipelines, embedding pipelines, or ML-serving datasets.
• Data quality engineering: validation, duplicate detection, missing-value handling.
• English reading proficiency for technical documentation.

Preferred Skills (strong plus)

• Iceberg REST catalogs (Lakekeeper, Polaris, or Nessie); OpenMetadata; Superset.
• Kafka operations and schema registry management.
• Running data workloads on Kubernetes.
• Formal feature stores (e.g., Feast) or NLP data preparation for AI/ML.
• Spark for batch processing.
• Experience in regulated or data-residency-constrained environments (e.g., Vietnam PDPL 91/2025, Decree 53/2022).

How We Work

• Architecture evolves through documented ADRs and a change-classification process: standard changes ship fast, architectural changes get review.
• Documentation-first culture: conclusions up front, self-contained specifications, explicit assumptions.