Engineer, AI Framework Software

Engineer, AI Framework Software (ML Accelerator Compiler)

About the Role

We are looking for an experienced AI Compiler Engineer to join our AI compiler and toolchain team. You will design, implement, and optimize compiler infrastructure for machine learning workloads running on heterogeneous and custom AI accelerators.

This role focuses on compiler optimization, hardware-aware code generation, performance engineering, and production deployment of real-world AI workloads. You will work closely with compiler, runtime, hardware architecture, and machine learning teams to improve performance, efficiency, and reliability across the end-to-end AI software stack.

The ideal candidate has strong C++ skills, solid compiler or systems experience, and hands-on experience optimizing software for modern CPUs, GPUs, NPUs, or other specialized accelerators.

Responsibilities

• Design, implement, and optimize compiler passes and transformations for AI workloads and heterogeneous accelerator backends.
• Develop IR-level and backend-level optimizations, including graph lowering, operator fusion, scheduling, memory planning, layout transformation, and code generation.
• Optimize compiler-generated code for performance, memory efficiency, and hardware utilization.
• Develop hardware-aware compiler optimizations for custom AI accelerators or other heterogeneous computing architectures.
• Analyze end-to-end model workloads and identify compiler and system-level performance bottlenecks.
• Debug functional correctness, numerical accuracy, compilation, and performance issues across compiler pipelines.
• Work closely with hardware architects and runtime engineers to enable new accelerator features and improve hardware/software co-design.
• Build benchmarking, profiling, testing, and automation infrastructure for compiler development and performance validation.
• Evaluate compiler optimizations using real-world AI models and workloads, and drive improvements from prototyping through production.
• Contribute to compiler architecture, technical design, code reviews, and engineering best practices.

Required Qualifications

• Master’s degree or higher in Computer Science, Computer Engineering, Electrical Engineering, or a related field.
• 3+ years of relevant industry or research experience in compiler, systems software, AI infrastructure, or performance engineering.
• Strong proficiency in modern C++ and experience developing production-quality software.
• Strong understanding of data structures, algorithms, and software engineering fundamentals.
• Solid knowledge of at least one or more of the following:
• Compiler architecture and optimization
• LLVM / MLIR or other compiler infrastructures
• Computer architecture
• Operating systems and runtime systems
• GPU/NPU programming or heterogeneous computing
• High-performance computing
• Hands-on experience developing or optimizing compiler passes, intermediate representations, code generation, or performance-critical systems.
• Experience analyzing and debugging complex performance and correctness issues in large codebases.
• Ability to independently drive technical problems from investigation and design through implementation, validation, and optimization.

Preferred Qualifications

• Hands-on experience with LLVM, MLIR, TVM, XLA, Triton, or similar compiler frameworks.
• Experience with AI/ML compiler stacks, including model graph lowering, operator fusion, scheduling, memory planning, or backend code generation.
• Experience targeting GPUs, NPUs, TPUs, or other custom accelerators.
• Experience with hardware-aware optimization, including memory hierarchy, data layout, tiling, vectorization, parallelism, and memory bandwidth optimization.
• Experience with machine learning frameworks and model formats such as PyTorch, ONNX, TensorFlow, or related ecosystems.
• Experience with performance profiling and benchmarking tools.
• Experience with quantization, mixed-precision computation, or numerical optimization for AI inference.
• Experience collaborating with hardware, runtime, kernel, or model teams on production AI systems.
• Contributions to compiler-related open-source projects, technical publications, or research projects are a plus.

What You Will Work On

• End-to-end AI compiler pipelines, from model frontends and graph representations to hardware-specific code generation.
• Compiler optimizations for modern ML workloads, including graph-level and kernel-level performance improvements.
• Hardware-aware scheduling, memory optimization, data layout transformation, and code generation.
• Performance tuning for large language models, vision models, multimodal models, and other production AI workloads.
• Compiler and runtime integration for next-generation AI accelerators.
• Production-quality compiler software used to deploy and optimize real customer workloads.

What We Value

• Strong curiosity about how software maps onto hardware.
• A performance-oriented mindset and attention to measurable results.
• Ability to debug complex problems across compiler, runtime, and hardware layers.
• Ownership of technical problems and ability to work independently.
• Strong communication and collaboration skills across compiler, runtime, hardware, and ML teams.
• Ability to learn new compiler technologies and accelerator architectures quickly.