Location:
CN-Shenzhen-HyQ
Shift:
Standard - 40 Hours (China)
Scheduled Weekly Hours:
40
Worker Type:
Permanent
Job Summary:
Design, build, and operate critical subsystems of the data platform. Take ownership of major components — streaming, batch, storage, or serving — drive them from design to production, and continuously improve their reliability and performance.
Job Duties:
Responsibilities
- Streaming Infrastructure: Kafka cluster operations (broker tuning, partition rebalancing, monitoring, disaster recovery); manage schema registry and Kafka Connect connectors
- Batch & Streaming Compute: Build and optimize Spark batch jobs and Flink/Spark Structured Streaming pipelines; contribute reusable job frameworks and tuning guides
- Storage & Lakehouse: Manage Iceberg tables (compaction, snapshot expiration, orphan file cleanup, schema evolution); operate MinIO at scale (lifecycle rules, tiering, performance tuning)
- Query & Serving: Deploy and operate Trino clusters (connector config, resource groups, query monitoring); manage StarRocks/ClickHouse clusters (sharding, replication, materialized views)
- Orchestration: Build and maintain Airflow/Dagster DAGs for platform operations; extend custom operators and sensors
- Platform Observability: Implement monitoring with OpenTelemetry, Prometheus, Grafana, and Loki across all stack layers; build dashboards and alerting rules
- Kubernetes Operations: Write Helm charts, manage operator lifecycles, configure resource quotas, node affinity, pod disruption budgets
- CI/CD & GitOps: Own ArgoCD application sets, Helm-based deployments, and promotion pipelines from dev to prod
- Participate in on-call rotation; write post-mortems and runbooks
- Mentor mid-level engineers through pairing, design discussions, and code reviews
Required Skills & Experience
- 6+ years in data/platform engineering or backend infrastructure
- Strong Kubernetes: Helm chart authoring, RBAC, network policies, storage (PV/PVC), operators
- Solid Kafka: topic design, consumer group management, offset management, monitoring lag, Kafka Connect, schema registry (Apicurio or Confluent)
- Solid Spark: Dataframe/Dataset API, Spark SQL, performance tuning, troubleshooting in production
- Working knowledge of Flink or Spark Structured Streaming for real-time pipelines
- Practical Iceberg experience: table maintenance, time-travel, catalog integration
- Hands-on Trino or Presto: connector configuration, query tuning, resource group management
- Experience with an OLAP engine: StarRocks, ClickHouse, or Doris — table design, ingestion pipelines, query optimization
- Proficient in Python and either Scala or Java
- Solid CI/CD and GitOps: ArgoCD or Flux, Helm, Docker
- Airflow or Dagster for pipeline orchestration
Nice to Have
- OpenShift-specific: SCC, Routes, ImageStreams, BuildConfigs
- dbt project experience for data transformation and modeling
- Data quality frameworks: Great Expectations, Soda, or Deequ
- Kafka Streams or ksqlDB for stream processing
- Exposure to DataHub/Atlas for data discovery and lineage
Company Introduction:
ITD SZ
港交所科技（深圳）有限公司，是2016年12月28日于深圳市前海自贸区成立的外商独资企业。
作为港交所的技术子公司，港交所科技（深圳）有限公司主要是为集团及其附属公司提供计算机软件、计算机硬件、信息系统、云存储、云计算、物联网和计算机网络的开发、技术服务、技术咨询、技术转让；经济信息咨询、企业管理咨询、商务信息咨询、商业信息咨询、信息系统设计、集成、运行维护；数据库管理、大数据分析；以承接服务外包方式提供系统应用管理和维护、信息技术支持管理、数据处理等信息技术和业务流程外包服务。
