Our client are a global semiconductor company based.
Responsibilities:
*Architect Scalable Pipelines: Design and optimize distributed CI/CD workflows for complex, multi-language builds (C++, Python, and kernel-level code).
*Hardware-in-the-Loop (HIL) Automation: Automate the lifecycle of physical/virtualized accelerator clusters to validate code against real-world silicon performance.
*GPU/Accelerator Orchestration: Manage and scale containerized environments using Kubernetes and Helm, optimized for high-bandwidth compute tasks.
*MLOps Enablement: Build the infrastructure that allows researchers to deploy models without managing underlying software complexity.
*System Telemetry: Implement advanced monitoring for GPU utilization, thermal performance, and throughput regressions.
Key Requirements:
*Programming: Python and Bash for scripting; familiarity with C/C++ for performance optimization.
*CI/CD: Proven experience with GitHub Actions, Buildkite, or GitLab CI.
*Infrastructure: Hands-on experience with Terraform and Ansible (Infrastructure as Code).
*Orchestration: Strong Kubernetes (k8s) and Docker/Container lifecycle skills.
Desirable Skills:
*Accelerated Compute: Knowledge of CUDA, OpenCL, SYCL, or ROCm.
*Compilers/Frameworks: Experience with Triton, MLIR, PyTorch, or JAX.
*Inference/Scaling: Familiarity with vLLM, Triton Inference Server, or Slurm.