We are seeking a AI Platform Engineer with approximately 6–12+ years of experience in the field. Builds and operates the shared AI platform infrastructure — the paved road every AI practice builds on top of, so no team reinvents deployment plumbing.
Responsibilities:
- Build and maintain shared AI platform infrastructure — compute provisioning, networking, IAM for AI workloads
- Own the internal tooling and templates practices use to deploy models/agents consistently
- Standardize CI/CD pipelines for AI workloads across practices, including shared AI evaluation platforms
- Manage platform-level cost governance and capacity planning across concurrent engagements
- Own platform security posture in partnership with AI Security Engineers
- Partner with MLOps/LLMOps Engineers on the boundary between platform and workload-specific operations
- Balance competing infrastructure requests from multiple practice leads
- Document platform capabilities clearly enough that practices can self-serve
- Forecast and justify platform spend to non-technical leadership
Requirements:
- 6–12+ yrs platform/infrastructure engineering, with 2+ yrs supporting AI/ML workloads specifically
- Deep cloud infrastructure expertise (IaC, Kubernetes, networking, IAM), including hosting vector/graph databases
- Experience building internal developer platforms/tooling, not just running infrastructure
- Experience integrating and operating managed AI/agentic platforms — Microsoft Azure AI Foundry, AWS Bedrock, and Google Vertex AI — alongside self-hosted open-source stacks as a good-to-have
- Familiarity with multi-tenant capacity planning and cost allocation
- Experience with platform-level security hardening
- Cross-practice stakeholder management — balances competing infra requests from multiple practice leads
- Cost/capacity planning literacy — can forecast and justify platform spend to non-technical leadership
- Documents platform capabilities clearly enough that practices can self-serve
- Collaborative — builds shared infrastructure without becoming a bottleneck
- Success metrics: platform uptime/reliability · cost per workload vs. budget · practice self-service adoption rate
