Compiler Engineer LLVM; -MCAR

Job Summary

Math Works has a hybrid work model that enables staff members to split their time between office and home. The hybrid model provides the advantage of having both in-person time with colleagues and flexible at-home life optimizations. Learn More:

Our group is responsible for the core technology used in Math Works code generation products. We are doing cutting-edge work on program analysis, language transformation, compiler optimization, and just-in-time compilation for simulation from high-level textual and graphical languages. We generate optimized code for embedded CPUs, GPUs, FPGAs, and other targets. Our optimizing compiler is actively developed by an elite team of talented programming language and compiler experts.

We are seeking a skilled compiler engineer with industrial-strength software engineering skills to help us improve our shared code generation infrastructure.

Math Works nurtures growth, appreciates inclusivity, encourages initiative, values teamwork, shares success, and rewards excellence.

Design and implement new features in our core code generation infrastructure using your object-oriented and compiler development skills.

Deliver features using test-driven development, writing unit tests, and fast iterations.

Contribute creative workflows and tools to be used by internal and external customers.

Participate in cross-team collaboration across multiple development teams.

• Demonstrated experience in/with compiler development
• A bachelor's degree and 3 years of professional work experience (or a master's degree, or equivalent experience) is required.

• MS + 2 years, or PhD in Computer Science or related technical field.
• Practical compiler or interpreter implementation experience.
• Demonstrated ability to dive into and work with large codebases.
• Strong abstraction, algorithmic, and software architecture skills.
• Strong communication skills.
• Experience with LLVM, MLIR, or other intermediate representations is strong plus.
• Experience writing optimizations for deep-learning code generation a is plus.