HPC/AI Solution Architect - SME

Brookfield Global Recruitment Services

United States

On-site

USD 150,000 - 175,000

Full time

14 days+

A complete application in a minute

Tailored resume and cover letter, ready to send.

Get past ATS filters

Job summary

A leading recruitment service is seeking a skilled HPC/AI Solution Architect to provide technical leadership in deploying and optimizing HPC and AI solutions. This role requires a strong educational background in Computer Science or Engineering, alongside significant expertise in advanced technologies and performance tuning. The candidate will work closely with engineering teams, mentor staff, and ensure the deployment of robust architectures that meet modern demands.

Qualifications

Experience with NVIDIA DGX systems and GPU topologies.
Proven skills in VMware security and disaster recovery solutions.
Ability to mentor and train engineering teams.

Responsibilities

Provide technical leadership for HPC/AI solutions.
Implement and optimize high-performance computing architectures.
Conduct performance tuning and security enhancements.

Skills

GPU-accelerated HPC clusters

Performance Tuning

Container Security

Disaster Recovery

Education

Bachelor's or Master's degree in Computer Science or Engineering

Job description

Direct message the job poster from Brookfield Global Recruitment Services

As an HPC/AI Solution Architect, you will provide technical leadership in designing, implementing, and optimizing high-performance computing (HPC) and artificial intelligence (AI) solutions.

Experience, Skills and Responsibilities required:

Hold Bachelor's or Master's degree in Computer Science, Engineering, or a related field.
Provide end-to-end pre-sales and post-sales design and deployment.
Serve as SME.
Mentor and Train Engineering Teams.

Design and deploy GPU-accelerated HPC clusters required:

NVIDIA DGX A100, HGX H100/H200/B100/B200 systems with NVSwitch and NVLink topologies, configure MIG for multi-tenant GPU slicing.
Mellanox HDR & NDR InfiniBand fabrics with OpenSM subnet managers, configure MTU 9000, Adaptive Routing, and PKey segmentation.
Slurm Workload Manager (v23.x) integration with GPU plugin, fair share scheduler, and Gang Scheduling for MPI jobs.
Use parallel I/O clients via MPI-IO over Lustre, tune striping across OSS nodes, and leverage client-side prefetch.
Implement caching layers with NVMe SSDs as scratch (“burst buffer”), managed by Open OnDemand’s burst buffer plugin.
Containerized workflows using K8s v1.24+, NVIDIA GPU Operator, DCGM exporter for telemetry, and Katib for hyperparameter tuning experiments.
Tune RDMA queue pair counts (mlx5_core parameters), increase credit values, and adjust Congestion Control settings.
Benchmark with fio (64 K sequential, depth 128), iozone, and NCCL all reduce tests, capturing host/end-to-end jitter metrics.

Performance Tuning & Security skills required:

ESXi host optimizations: CPU Ready Time reductions, NUMA alignment, storage multipathing PSP selection.
Storage array tuning: queue depth adjustments, cache reservation, LDEV alignment to 1 MB boundaries.
Full-disk encryption: LUKS2 with Argon2id key derivation on Linux, FIPS 140-2 validated modules, TPM 2.0 sealing.
VMware security: Secure Boot, UEFI lockdown, vTPM, VM encryption with vSAN encryption and Key Provider integration (Cisco ISE PKI).
Container security: PodSecurityPolicy/OPA Gatekeeper enforcement, SELinux/AppArmor in enforcing mode, Cilium for L3-7 network policies.
Identity and access: LDAP/AD integration over LDAPS, SAML 2.0/OAuth2 SSO for portal access, Role-Based Access Control (RBAC) in Kubernetes and vCenter.

Disaster recovery and continuity skills required:

Define RTO/RPO with business units, implement stretched clusters (vSAN stretched, Ceph multisite).
Automated failover orchestration via Ansible Playbooks and Terraform for IaC-driven recovery procedures.

Seniority level