Note: The job is a remote job and is open to candidates in USA. AxeGenAI is standing up production AI Factory infrastructure for an enterprise client. The Principal AI Factory Engineer will lead NVIDIA H200 cluster deployment, build a multi-tenant inference platform with token metering and observability, and deliver production GenAI applications including financial-domain RAG and NLP pipelines.
Responsibilities
- Lead bring-up of NVIDIA HGX/DGX H200 systems: racking, power, cooling, InfiniBand/NDR fabric, NVLink, firmware/Redfish, drivers and CUDA
- Run cluster validation and burn-in (NCCL, GPU diagnostics, thermal/power soak)
- Stand up orchestration: Kubernetes + NVIDIA GPU Operator (or Slurm)
- Build the serving layer: vLLM / Triton / NVIDIA NIM, API gateway, autoscaling
- Implement token metering, quota and multi-tenant isolation, usage billing hooks, and observability
- Ship production GenAI use cases: a financial-domain RAG assistant (retrieval, vector store, embeddings, guardrails), sentiment/NLP pipelines, and more
- Build the data pipelines and evals these workloads depend on
Skills
- Hands-on H100/H200 cluster deployment — the metal, not just the guide
- Strong MLOps: containerized GPU workloads, K8s, production model serving at scale with real multi-tenancy
- Proven delivery of production LLM/RAG applications — shipped, not demoed
- InfiniBand/RDMA, Linux/BMC fundamentals, strong Python and modern GenAI tooling
- The judgment to sequence three disciplines under a tight timeline and spot gaps before they bite
- Remote coordination with on-site deployment windows (international travel may be required)
- NVIDIA certifications (NCP/NCA), liquid-cooled deployment experience
- Telecom, financial services, or hyperscale exposure
- Token metering / GPUaaS billing experience
Benefits
- Remote coordination, with on-site deployment windows
Company Overview