Senior Data Platform Engineer

Senior Data Platform Engineer — Cloudera, AWS & Kubernetes

The mission

We are looking for a battle-tested Data Platform Engineer who can build, operate, troubleshoot, and evolve large-scale data platforms across on-premises Cloudera environments and cloud-native AWS/Kubernetes architectures.

This is not a dashboard or SQL-only role. You will work where distributed compute, storage, networking, Kubernetes, and production data pipelines meet. You must be comfortable tracing a failed workload from the application layer through Spark or Flink,

Kubernetes operators, HDFS/Hive, infrastructure, and AWS services.

The goal is to help us evolve safely from on-premises, VM, and EC2-based platforms into resilient, observable, cloud-native data systems.

What you'll do

- Build and operate production data platforms across Cloudera on premises, Cloudera cloud environments, AWS EMR, and Kubernetes.

- Design batch and streaming pipelines using Spark, Flink, Hive, and related technologies.

- Transform large datasets through filtering, sorting, joining, aggregation, partitioning, enrichment, and restructuring.

- Work with Parquet, Avro, JSON, CSV, and other delimited or semi-structured formats.

- Design storage, partitioning, compression, retention, and lifecycle strategies across HDFS, Hive, and object storage.

- Design and maintain Hive schemas, tables, partitions, metadata, and data models.

- Administer Cloudera clusters, including installation, upgrades, configuration, scaling, patching, security, backup, and recovery.

- Troubleshoot unhealthy services, failed jobs, resource contention, data skew, small-file problems, metadata issues, and storage bottlenecks.

- Tune Spark and Flink workloads for memory, CPU, parallelism, shuffle behavior, checkpointing, and recovery.

- Operate AWS services such as EMR, S3, IAM, EC2, EKS, CloudWatch, KMS, and supporting networking services.

- Deploy and operate data workloads on Kubernetes using operators, Helm, custom resources, and GitOps-based delivery.

- Help migrate workloads from physical or virtual machines and EC2 into AWS and Kubernetes-based platforms.

- Separate compute from storage where appropriate while accounting for performance, resilience, security, and cost.

- Build monitoring, alerting, capacity management, and operational runbooks for critical data services.

- Automate platform provisioning and configuration using Terraform, Ansible, scripting, and CI/CD.

- Support production incidents involving failed pipelines, delayed data, cluster degradation, storage pressure, or infrastructure failure.

- Work with data engineering, infrastructure, security, and application teams to resolve problems across ownership boundaries.

What we're looking for

- Strong hands-on experience administering Cloudera platforms in production.

- Experience with both on-premises Cloudera and cloud-based Cloudera deployments.

- Deep working knowledge of Hadoop, HDFS, Hive, YARN, Spark, and the wider distributed-data ecosystem.

- Experience building or operating production workloads using Apache Flink.

- Strong understanding of distributed data processing, including partitioning, shuffling, serialization, checkpointing, and failure recovery.

- Experience transforming large datasets using joins, aggregations, filtering, sorting, and schema evolution.

- Practical knowledge of Parquet, Avro, JSON, CSV, compression formats, and serialization tradeoffs.

- Experience designing data layouts for query performance, ingestion throughput, retention, and cost.

- Strong AWS experience, particularly with EMR, S3, EC2, EKS, IAM, CloudWatch, and KMS.

- Strong Kubernetes experience, including operators, controllers, Helm, scheduling, storage, networking, and workload troubleshooting.

- Experience migrating data platforms from on-premises or VM-based environments into AWS and Kubernetes.

- Ability to troubleshoot Linux, JVM, networking, storage, DNS, certificates, and resource-management issues.

- Experience with observability platforms and the ability to correlate infrastructure symptoms with data-pipeline failures.

- Ability to automate operational work using Python, Bash, Terraform, Ansible, or equivalent tools.

- Strong judgment around production changes, data integrity, access control, rollback, and recovery.

Production scenarios you should be able to handle

- A Spark job that ran in 40 minutes yesterday now takes four hours.

- A join creates severe data skew and repeatedly exhausts executor memory.

- HDFS is approaching capacity while NameNode health is degrading.

- Hive queries return incomplete results because partitions or metadata are inconsistent.

- A Flink job repeatedly fails after checkpoint recovery.

- An EMR workload is reliable but significantly more expensive than expected.

- A Kubernetes operator reports success while the underlying data workload is unhealthy.

- A migrated workload behaves differently on S3 than it did on HDFS.

- A certificate, Kerberos, IAM, DNS, or network problem presents as an application failure.

- A critical pipeline misses its SLA and ownership is unclear across platform and data teams.

Certifications

Relevant certifications are useful, particularly:

- Cloudera Certified Professional: Data Engineer

- Cloudera Administrator certification or equivalent production experience

- AWS data, analytics, or architecture certifications

- Kubernetes certifications such as CKA or CKAD

What success looks like

- Data pipelines meet their reliability and processing-time objectives.

- Platform failures are detected before downstream consumers report them.

- Incidents move quickly from symptoms to an evidence-backed root cause.

- Cloudera, AWS, and Kubernetes environments are operated through repeatable automation.

- Migrations preserve data correctness while improving scalability and operability.

- Storage and compute designs balance performance, resilience, and cost.

- Data engineers can ship workloads without becoming accidental platform administrators.

- Operational knowledge becomes monitoring, automation, and runbooks—not tribal memory.

The person we want

You understand that a data platform is a distributed production system, not a collection of product names.

You can move from a Hive execution plan to Spark executor logs, Kubernetes events, HDFS health, S3 behavior, IAM permissions, and network telemetry without losing the thread. You know the architectural differences between on-premises Hadoop and cloud-

native data platforms, including where a lift-and-shift approach will fail.

We need someone who can enter a degraded platform, establish the facts, protect data integrity, restore service, explain the failure chain, and make the system harder to break next time.