Remote Mojo Compiler Engineer — AI Language & GPUs

Modular; United States; Remote; USD 135,000 - 242,000; Full time

Benefits offered by this job

Amazing Team

World-class Benefits

Competitive Compensation

Team Building Events

Job summary

Modular is seeking a Compiler Engineer to design and evolve the Mojo language compiler, built on MLIR, enabling high-performance heterogeneous compute for AI systems. You will work across GPUs, CPUs, and accelerators to push the boundaries of language design and toolchains.

You will collaborate with core teams, researchers, and an open-source community while shaping the Mojo developer experience from syntax to performance semantics.

Qualifications

• 3+ years on compilers, runtimes, or language implementations.
• Hands-on mastery of C++; Python fluency a big plus.
• Familiarity with MLIR and LLVM frameworks.
• Understanding of type systems, compiler optimization theory, and language semantics.
• Contributions to real-world language implementations (Clang, GCC, Swift, Rust, Dart).

Responsibilities

• Craft and implement new Mojo language features on top our MLIR-based compiler architecture.
• Dream up and realize optimization passes that make GPUs, CPUs, and accelerators sing in harmony.
• Shape the developer experience for Mojo – from syntax to performance semantics.
• Collaborate with the Modular core teams, researchers, and an emerging open-source community.
• Help define what "next-generation language design" means in the age of AI.

Skills

C++ mastery

Python fluency

MLIR/LLVM

Language semantics

Compiler optimization

Tools

MLIR

LLVM

Clang

Job description

Modular is seeking a Compiler Engineer to design and evolve the Mojo language compiler, built on MLIR, enabling high-performance heterogeneous compute for AI systems. You will work across GPUs, CPUs, and accelerators to push the boundaries of language design and toolchains.

You will collaborate with core teams, researchers, and an open-source community while shaping the Mojo developer experience from syntax to performance semantics.