High-Performance Computing DevOps Architect

Our Team

Our team is developing a high-performance computing solution for low-latency and high throughput image processing and deep-learning workloads that will enable our Chip Manufacturing process control equipment to offer differentiated value to our customers.

Your Opportunity

As an HPC Architect you will get the opportunity to architect high-performance computing solutions from scratch and design/optimize all aspects (Compute Memory Networking Storage) for better cost of Ownership.

Roles and Responsibility

• As an architect you will be responsible for designing HPC infrastructure solutions including compute networking storage and workload management components.

• You will work closely with cross-functional teams including Hardware Software product management and business stakeholders to understand compute workload and translate them into Platform architecture and designs that meet business needs.

• You will create and maintain detailed system architecture diagrams and specifications.

• You will evaluate and select appropriate hardware and software components for HPC environments

• You will Install configure and maintain HPC systems including hardware software and networking components

• You will develop and implement automation scripts for system management and deployment.

• You will be a subject Matter expert to unblock dependent teams in the HPC domain.

• You will be expected to develop system benchmarks profile systems to understand bottlenecks optimize workflows and processes to improve cost of ownership.

• Identify and mitigate technical risks and issues throughout the HPC development life cycle.

• Ensure that Compute Cluster is resilient reliable and maintainable.

• You will be expected to stay abreast of the latest HPC technologies including Hardware Software and Networking Solutions

• Your primary focus will be to understand the compute workload and design HPC cluster with right combination of Nodes CPU/GPU Memory Interconnects and storage to have optimum performance at minimum cost of Ownership.

Our Ideal Candidate

Someone who has the drive and passion to learn quickly has the ability to multi-task and switch contexts based on business needs.

Qualifications

• In-depth experience with Linux System administration and Hardware/Software Configuration.

• Strong knowledge of HPC technologies including cluster computing high speed interconnects (InfiniBand RoCE) parallel filesystems (Lustre GPFS BeeGFS etc)

• Experience in creating maintaining Operating System images with different installation and boot schemes

• Extremely good with automation tools like Ansible Chef Salt-Stack and Scripting languages (Python and Bash)

• Experience in Creating maintaining Storage Solutions with different RAID configuration.

• Ability to design storage solution for different IOPS Access patterns (Random vs Sequential RW) and tune storage and filesystems for better performance.

• Good of knowledge Networking concepts including IP addressing routing protocols and Switch configuration for RDMA VLAN configuration network bonding etc.

• Good Knowledge Virtualization Hardware and Software Hypervisors

• Good knowledge of containerization technologies like docker singularity.

• Experience in Software Defined Networking and Storage.

• Experience in setting-up remote management protocols like IPMI Redfish etc.

• Experience in setting-up and using monitoring systems like Prometheus Grafana.

• Experience System profiling and custom tuning for target workload for higher performance and low cost of ownership

• Very good written and verbal communication skills.

• Very good in Technical documentation meant to serve as manuals for non-experts in the field.

Additional Qualifications:

• Experience in HPC Cluster management and Work-load orchestration software (e.g. SLURM Torque LSF)

• Experience in Setting-up Deep-learning training/inference solutions.

• Experience in Private cloud infrastructure like Kubernetes OpenStack CloudStack etc.

• Experience in Distributed High Performance Computing and Parallel programming frameworks

• Good knowledge of Low-latency and high-throughput data transfer technologies (RDMA on RoCE InfiniBand)