# GPU Cluster Engineer, Systems & Platform — Sciforium

- Location: San Francisco, CA
- Posted pay: $150k–$220k/yr
- Posted: Aug 6, 2026
- Status: open (checked Nov 20, 2026)
- Apply (employer's site): https://jobs.ashbyhq.com/Sciforium/62d7ad5e-d00b-4e6d-b5be-b02ea9ef65ea
- Employer record: https://visajobs.io/company/sciforium.md

## Sponsorship history
- Tier: Employer history only
- With this exact title: 0

## Description
Sciforium is an AI infrastructure company developing next-generation multimodal AI models and a proprietary, high-efficiency serving platform. Backed by multi-million-dollar funding and direct sponsorship from AMD with hands-on support from AMD engineers the team is scaling rapidly to build the full stack powering frontier AI models and real-time applications.
About the Role
We are looking for a GPU Cluster Engineer to own the entire software stack of our GPU clusters — from kernel tuning and GPU drivers up through schedulers, containers, and ML frameworks. While our Hardware Operations team keeps the physical machines healthy and connected, you define what a production-ready node looks like in software: you author the images, playbooks, and pipelines that take a freshly provisioned server to a fully validated GPU node, and you keep the fleet consistent, upgradable, and fast. You will serve two demanding customer groups — our foundation model training teams and our model serving/product teams — ensuring both run on correctly configured, well-managed, high-performance infrastructure.
Key Responsibilities
OS Bring-Up & Node Lifecycle Engineering
Golden Images & Automated Bring-Up: Own the node software definition — versioned OS images, kernel tuning (NUMA, hugepages, IRQ affinity, cgroups), GPU/NIC driver stacks — and the automated pipeline that takes a node from base OS to production-ready.
Validation & Burn-In: Build automated acceptance suites (DCGM diagnostics, nccl-tests/RCCL tests, bandwidth and topology checks, HPL) that gate every node before it enters a scheduler pool.
Fleet Maintenance: Execute rolling kernel/driver/toolkit upgrades with minimal disruption to running workloads; enforce configuration consistency, detect drift, and maintain the driver ↔ CUDA/ROCm ↔ framework compatibility matrix across the fleet.
Self-Healing Operations: Automate detection of unhealthy nodes (Xid/ECC errors, link flaps, thermal throttling), with cordon/drain/reboot/re-image workflows and clean handoff to Hardware Operations for physical repair or RMA.
Configuration Management & Automation
Infrastructure as Code: Manage all node and cluster configuration through Ansible/SaltStack playbooks in Git, with peer-reviewed changes, CI validation, and canary rollouts before fleet-wide deployment.
Provisioning Pipelines: Build and maintain image/provisioning tooling (PXE, MaaS, Packer, or similar) so new or re-imaged nodes are reproducible, not hand-crafted.
Operational Tooling: Develop Python/Bash tooling for cluster operations, health reporting, and workflow automation.
Orchestration & Scheduling (Kubernetes & Slurm)
Kubernetes for Serving: Deploy and operate GPU-enabled Kubernetes for inference workloads — NVIDIA GPU Operator, device plugins, node feature discovery, topology-aware scheduling, and MIG/MPS partitioning where appropriate.
Training Schedulers: Operate Slurm (or Run:AI) for multi-node training — partitions, QoS, preemption, accounting, and container integration (enroot/pyxis).
Container Platform: Maintain base images, registries, and the NVIDIA Container Toolkit / ROCm container stack; keep training and serving images lean, current, and reproducible.
GPU Driver & ML Stack Engineering
Driver & Runtime Lifecycle: Build, deploy, and debug the full accelerator stack — NVIDIA (CUDA toolkit, cuDNN, NCCL, Fabric Manager) and AMD (ROCm, RCCL) — including kernel modules (DKMS), GPUDirect RDMA/Storage, and the RDMA software stack (MOFED/DOCA).
Framework Environments: Maintain curated, optimized PyTorch and JAX environments with sane dependency and version management for researchers and production services.
Distributed Performance: Tune NCCL/RCCL across NVLink/NVSwitch and InfiniBand/RoCE fabrics, ensure topology-aware job placement, and run continuous communication/throughput benchmarks to catch regressions.
Advanced Debugging & Observability
Escalation Point: Own the hard problems — NCCL hangs and timeouts, CUDA memory leaks, ROCm kernel crashes, straggler nodes, and unexplained throughput drops.
Observability: Own software-layer monitoring (DCGM exporter, Prometheus/Grafana, alerting) plus job-level GPU utilization and cluster efficiency reporting.
Qualifications
Must-Haves:
5+ years in systems/infrastructure engineering with significant GPU cluster, HPC, or large-scale ML infrastructure experience.
Bachelor's or Master's degree in Computer Science, Computer Engineering, Electrical Engineering, or a related technical field.
Deep Linux internals expertise: kernel modules/DKMS, systemd, cgroups, NUMA, and system performance tuning.
Hands-on experience with NVIDIA (CUDA) and/or AMD (ROCm) driver and runtime stacks on modern accelerators (H200/B200, MI325x/MI355x class), including kernel-level debugging.
Production Kubernetes experience with GPU workloads, plus working knowledge of HPC schedulers (Slurm/Run:AI) — or the reverse (deep Slurm, working K8s).
Strong configuration management experience (Ansible or SaltStack) with Git-based, code-reviewed infrastructure workflows.
Provisioning and image tooling experience (Packer, MaaS, Foreman, Terraform, or similar) for automated, reproducible node builds.
Client-side experience with distributed filesystems (Lustre, GPFS, Weka) and checkpoint I/O optimization.
Container fluency: Docker/containerd and the NVIDIA Container Toolkit or ROCm equivalent.
Proficiency in Python and Bash for automation and tooling.
Working knowledge of NCCL and RDMA networking (InfiniBand/RoCE, GPUDirect) and of PyTorch/JAX runtime behavior.
Nice-to-Haves:
Experience directly supporting foundation model training teams — multi-node job failure debugging, checkpoint pipeline tuning, and framework-level performance triage — ideally in a startup or research-heavy environment.
Experience deploying and tuning inference/serving stacks (vLLM, Triton Inference Server, TensorRT-LLM) for latency and throughput targets.
GPU/system profiling tools: Nsight Systems/Compute, rocprof, perf, eBPF.
Benef

> History, not a promise: past H-1B filings do not mean an employer will sponsor you or a given role. Data: U.S. DOL OFLC LCA disclosure (FY2024–FY2026 for role counts, certified LCAs) and the USCIS H-1B Employer Data Hub. Not legal advice. Methodology: https://visajobs.io/methodology
