Staff Software Engineer, Compiler

Code Metal is developing translation tools that convert Python algorithm code into correct, deployable C++ for embedded systems. The Staff Software Engineer will lead the translation pipeline architecture, from parsing and semantic analysis through IR design and C++ code generation, while integrating AI assistance, testing, and verification.

Responsibilities:

• Design and evolve the pipeline's IR, analysis passes and C++ code generation, keeping the core independent of any one source language
• Build type inference and lowering for dynamic Python features: arbitrary-precision integers, exceptions, generators, containers, and calls to the standard library and third-party libraries
• Make semantic equivalence measurable, through differential testing against CPython, conformance suites and fuzzing
• Integrate LLM-assisted generation where analysis falls short, with structured outputs and checks that catch its mistakes
• Collaborate with verification and domain specialists to bring static analysis and formal methods into the pipeline
• Set engineering standards for the codebase: tests, CI/CD, reproducible builds and code review

Requirements:

• 7+ years of professional software development in C++ and/or Python, shipping to real users
• Hands-on experience building a compiler, transpiler, interpreter or language runtime: the end-to-end pipeline, not just one pass
• Deep understanding of parsing, AST and IR design, type systems, IR transformations and code generation
• Experience with compiler infrastructure such as Clang/LLVM, MLIR, GCC or equivalent
• Strong modern C++: templates, value semantics, undefined behaviour, and how source code maps to the machine
• Solid delivery fundamentals: code review, testing, CI/CD and build systems
• A track record of leading large technical projects in a team, and of turning stakeholder needs into designs
• Eligible to obtain and maintain an active U.S. Top Secret security clearance
• Experience with dynamic-language implementation or analysis (CPython internals, PyPy, Cython, Numba, mypyc, or type inference for dynamic languages)
• Differential testing, fuzzing, symbolic execution, or formal verification (e.g. Z3, Coq, TLA+)
• Experience applying LLMs to code generation or program repair
• Embedded, heterogeneous (FPGA/GPU) or resource-constrained targets
• Familiarity with DSP, RF, communications or other compute-intensive domains
• A current security clearance, or the ability to obtain one (US citizenship required for cleared work)