Cloud Compute Engineer

Castelion Corporation•Torrance, CA

About The Position

Castelion is seeking a Cloud Compute Engineer to build and own the Seeker Team’s scalable cloud compute platform for data processing, simulation, calibration, and machine learning. The role will move compute-intensive workflows from individual laptops and one-off scripts to a shared, automated, reliable platform used by multiple engineering teams. Working across data processing, ML training, and test infrastructure, the engineer will build distributed compute systems, optimize Python workloads, and integrate simulation and regression testing into CI/CD pipelines. The position is hands-on and will affect engineering productivity, data-processing speed, and the scalability of seeker development. The role also involves working with engineers to establish consistent compute patterns across unclassified and classified networks. Access to classified information or restricted U.S. Government sites, systems, or information may be required depending on the position’s designation. If applicable, employment may be contingent on obtaining and maintaining required U.S. Government security clearance or other authorization and meeting applicable citizenship or eligibility requirements. The position requires access to export-controlled information, so employment is contingent on the candidate being able to access applicable information without additional export licensing. Castelion is an Equal Opportunity Employer and provides reasonable accommodations to qualified individuals with disabilities during the application and hiring process. Duties and responsibilities may change at any time.

Requirements

Experience building and operating distributed compute platforms used by multiple engineering teams.
Strong Python skills, including profiling, concurrency, memory management, and performance optimization.
Production experience with Ray and Kubernetes.
Experience designing batch-processing, event-driven, and user-submitted job systems.
Experience with Azure compute, storage, networking, identity, and monitoring services.
Demonstrated ability to optimize application code and data pipelines for performance, reliability, and cost.
Experience implementing observability, automated job recovery, and reproducible processing workflows.
Ability to build simple developer interfaces and work directly with users to automate complex workflows.
Ability to access applicable export-controlled information without additional export licensing. If designated, ability to obtain and maintain required U.S. Government security clearance or other authorization and satisfy applicable citizenship or eligibility requirements.

Nice To Haves

Strong PyTorch experience, including distributed training, data loading, checkpointing, and GPU utilization.
Experience scaling ML training across large datasets and distributed compute resources.
Experience integrating simulation and regression workloads into CI/CD pipelines.
Familiarity with large-scale engineering datasets, flight-test data, calibration, or scientific computing workflows.
Experience building compute platforms that support both classified and unclassified environments.
Experience improving processing throughput and turnaround time through workload optimization and infrastructure scaling.

Responsibilities

Build and operate distributed compute infrastructure for batch processing, custom jobs, simulation, calibration, preprocessing, and ML training.
Build event-driven workflows that automatically process newly ingested data, including flight-test and chamber calibration data.
Enable scalable ML training, Foxglove RDL preprocessing, data preparation, replay, and other user-submitted jobs.
Integrate simulation and regression tests into CI/CD pipelines to expand test coverage beyond standard runner limitations.
Profile and improve Python workloads, data flows, memory usage, concurrency, and resource utilization to optimize performance and cost.
Provide interfaces, APIs, or SDKs that let engineers submit, monitor, debug, and rerun jobs without manually managing infrastructure.
Implement logging, metrics, alerting, job traceability, and cost monitoring to support failure diagnosis and recovery.
Track code versions, inputs, configurations, and outputs so compute jobs can be reproduced and audited.
Design efficient approaches to storage, data transfer, caching, partitioning, and preprocessing for large-scale data movement.
Work with engineers to understand their workflows and establish consistent compute patterns across unclassified and classified networks.