Compiler Engineer (GPU/Python)

Location: Hybrid in Paris or London

Company

hirify.global develops Quadrants, an open-source translation layer that turns ordinary Python into optimized machine code for physics and performance-critical numerical computing.

What you will do

• Own Quadrants end to end, including its API, intermediate representation, lowering, optimization passes, numerical primitives, and roadmap.
• Design an idiomatic Python authoring experience with clean decorators, legible constraints, clear errors, and lower-level escape hatches.
• Build and extend IR, lowering, and backend support across CUDA, x86, Arm64, AMDGPU, Metal, and Vulkan.
• Optimize generated kernels through vectorization, register and shared-memory usage, barriers, atomics, cooperative threading, and reduced synchronization.
• Develop and maintain numerical primitives such as Cholesky, eigendecomposition, matrix multiplication, reductions, and solver kernels.
• Benchmark real Genesis and Quadrants workloads, track performance regressions, improve abstractions, and support adoption by external users.

Requirements

• Strong GPU programming knowledge, including access patterns, registers, shared memory, barriers, atomics, and threading.
• Experience with idiomatic Python and designing APIs for Python users.
• Understanding of compiler pipelines from authoring through IR, lowering, optimization, and kernel generation.
• Ability to reason about whole systems, measure before optimizing, and prioritize memory traffic and launch overhead.
• Proactive, self-directed working style and comfort designing an open-source project for external users.
• Readiness to grow into leading a small team.

Nice to have

• CPU performance knowledge, especially for x86 and real-time simulation.
• Experience with inline assembly, C, or PTX escape hatches.
• Interest in differentiable simulation and performance-critical numerical computing.

Culture & Benefits

• Work on a standalone open-source project with its own roadmap.
• High ownership and autonomy, with responsibility expanding over time.
• Close collaboration with engineers using Quadrants, including Genesis users.
• Open design and development focused on readable code, maintainability, and measurable performance.
• AI agents may be used as productivity tools, while engineers retain code ownership.