AI Infrastructure Engineer

The AI Infrastructure Engineer is a platform specialist responsible for architecting, building, and operating high-performance AI infrastructure to support advanced AI workloads, including LLMs, GenAI, Computer Vision, and MLOps. This role will focus on managing GPU clusters (NVIDIA A100/H100), deploying and maintaining Red Hat OpenShift AI (RHODS), and ensuring secure, scalable, and cost-efficient AI platforms across SDD’s Sovereign Cloud and hybrid/multi-cloud environments. The engineer will enable enterprise-grade AI adoption for 200+ government entities.

Key Responsibilities & Deliverables

GPU & AI Platform Architecture

Design and implement GPU-based compute clusters. Define reference architectures for LLM hosting, Vector Databases, MLOps, and high-performance storage/networking.

Fully operational GPU-based AI infrastructure. GPU Cluster Uptime and Performance Utilization. Reduction in Cost per Training/Inference Workload.

GPU Cluster Operations

Install, configure, and optimize core components: CUDA, cuDNN, NCCL, NVIDIA Drivers, and GPU Operators. Implement GPU partitioning, scheduling, and performance tuning for high-end GPUs (e.g., A100/H100).

High-availability architecture for all AI workloads. Complete documentation and runbooks.

OpenShift AI (RHODS) Management

Deploy, configure, and maintain the Red Hat OpenShift AI (RHODS) platform for multi-tenant use. Manage the integration of NVIDIA GPU Operator for efficient GPU scheduling and support Data Scientists with Notebooks, Training, and Inference Endpoints.

Production-ready OpenShift AI (RHODS) platform. AI Project Onboarding Speed.

LLM & Model Serving

Build and manage infrastructure for hosting and serving open-source LLM frameworks (Llama, Falcon, Mistral) and supporting RAG pipelines, LoRA adapters, and Vector Databases (Milvus, pgvector).

Multi-model LLM serving environment for entities. MLOps Pipeline Success Rate and Deployment Frequency.

MLOps & Automation

Implement IaC (Terraform, Ansible) and GitOps for the automated lifecycle management of the AI platform (node onboarding, scaling, model rollout/rollback). Build robust MLOps pipelines for data prep, training, evaluation, and monitoring (using tools like MLflow/Kubeflow).

Infrastructure automation via Terraform & Ansible. Automation Coverage for AI Infrastructure.

Required Qualifications & Experience

• Experience: 7–12 years in Cloud Infrastructure, DevOps, ML Infrastructure, or Platform Engineering.
• Deep Hands-On Expertise:
• GPU Systems (NVIDIA A100/H100), Linux, Containers, and Kubernetes.
• OpenShift AI (RHODS) or equivalent Kubernetes GPU orchestration.
• LLM Hosting (Llama, Mistral, Falcon, etc.) and supporting Vector Databases/RAG systems.
• Strong Experience In: TensorFlow, PyTorch, Hugging Face, Distributed Training (DDP, Deep Speed), and ML Ops Stacks (ML flow, Kubeflow).

Essential Skills & Competencies

• Technical: Deep understanding of GPU compute, HPC architectures, and ML performance profiling. Strong skills in IaC (Terraform/Ansible), CI/CD, and OpenShift/Kubernetes operators.
• Soft Skills: Strong troubleshooting, optimization, and performance engineering mindset. Excellent cross-functional collaboration and documentation skills.

Preferred Certifications

• NVIDIA Deep Learning / AI Infrastructure Certification
• Red Hat OpenShift AI specialization
• Kubernetes CKA/CKAD
• Azure AI or Oracle Cloud AI certifications
• Terraform & Ansible certifications