About the job
In this role you will work closely with deep learning compiler engineers to build the infrastructure and automation that powers day-to-day development and releases. Responsibilities include designing and maintaining sophisticated CI/CD systems that run ML workloads at scale across diverse GPU environments, produce actionable signals for compiler developers, testers, and release engineers, and continuously improve stability and turnaround time. This includes building performance-aware pipelines and workload harnesses that support release confidence and long-term quality of deep learning compiler stacks.
Responsibilities
Drive CI and infrastructure capabilities that make deep learning compiler development fast, reliable, and scalable. This includes improving signal-to-noise (flake reduction, reproducibility, and richer diagnostics), accelerating iteration cycles, scaling capacity and coverage across models/hardware/software configurations, and building strong observability (metrics, logging, tracing, dashboards) so failures are easy to understand and fix.
Explore practical uses of AI to enhance CI workflows—such as smarter test selection, automated triage/summarization, and faster issue isolation—ultimately increasing the quality and speed of deep learning compiler development, testing, and release.
Qualifications
Minimum
BS, MS, or PhD (or equivalent experience) in Computer Science, Computer/Electrical Engineering, Mathematics, or related field
3+ years of professional experience designing and scaling CI/CD, build/release, or developer productivity infrastructure for DL/GPU software environments
Strong software engineering skills (Python required) with ability to architect, implement, and debug complex systems end-to-end
Hands-on experience building CI/MLOps platform capabilities—pipeline orchestration, artifact/package management, and production-grade observability (logs/metrics/dashboards)—with strong reliability and maintainability
Experience with deep learning frameworks/runtime stacks (e.g., PyTorch, JAX, vLLM, SGLang, TensorRT, NeMo) and running real workloads in production-like environments
Working knowledge of Linux-based development and debugging across complex software/hardware stacks (drivers, CUDA libraries, containers, cluster schedulers, etc.)
Preferred
Experience applying AI/LLMs and agent-based workflows to improve CI and infrastructure (e.g., smarter triage/routing, automated failure summarization, intelligent test selection, regression isolation, or developer-assist tooling)
Experience with compiler-focused verification techniques (e.g., differential testing across backends/versions, IR-level checks, automated reduction/minimization, fuzzing/property-based testing, or translation-validation style approaches)
Compiler-adjacent knowledge, including familiarity with LLVM/MLIR-based toolchains and the ability to debug issues that span compilation/codegen, runtime execution, and hardware/software boundaries