Principal AI Compiler Engineer

About the role

NXP is looking for a hands-on AI Compiler Engineer who works at the intersection of AI, compiler technology and hardware design. The role involves architecting and scaling a production-class AI compiler toolchain that maps neural networks onto NXP's SoCs, working with both human and AI engineers to build adaptive compilers. Locations available: San Diego, San Jose, California or Austin, Texas.

Key responsibilities

• Own the design, implementation and evolution of an AI compiler toolchain that uses AI agents to map neural networks onto NXP's SoC platforms.
• Pioneer graph transformations, lowering, scheduling and code generation strategies for CPUs and custom accelerators, guided by AI-powered analytics.
• Integrate deeply with AI frameworks such as PyTorch, TensorFlow and ONNX, using AI agents to onboard new model architectures and operations quickly.
• Advance quantization, operator fusion, memory planning and layout transformations, combining human expertise with AI-guided design.
• Partner with hardware and software architects, kernel hackers and AI agents to co-design compiler and accelerator features.
• Diagnose performance bottlenecks with AI-enabled profiling and tune for latency, throughput and power efficiency.
• Improve validation, benchmarking and regression pipelines with AI agents to ensure correctness and performance.
• Improve the developer experience by streamlining usability, diagnostics and documentation, with AI agents supporting user support and troubleshooting.

Required profile

• MS/PhD (or equivalent experience) in Computer Science, Electrical Engineering or a related field.
• Deep experience building AI compilers, accelerator backends or graph optimization frameworks.
• Strong expertise in graph optimization and performance optimization for NPUs or custom accelerators.
• Solid understanding of AI inference workloads such as CNNs, transformers and generative models.
• Excellent communication skills, including agile development experience in a Scrum team (e.g. Product Owner or Scrum Master).

Required skills

• MLIR, LLVM, TVM-like systems or proprietary compiler IRs
• C/C++ and Python
• Graph optimization and performance tuning