Senior Software Compute Architect (f/m/d)

In this role you will own the software architecture for a next‑gen hybrid electronic/photonic compute engine. You will define programming models, compilers, and runtimes to expose the hybrid hardware to AI frameworks and higher‑level libraries. You will evaluate AI workloads and guide system decisions to optimize performance across compute paths. Join a human‑centric, innovation‑driven team that aims to transform chip interconnects and compute efficiency. This opportunity offers impact across software–hardware co‑design, with a clear path to shaping the technology and its applications.

• comprehensive health insurance
• pension
• virtual stock options
• focus on work–life harmony
• autonomy to make a lasting impact
• professional and personal development

• Define programming model, instruction abstractions, and interfaces to expose hybrid compute to compilers, libraries, and AI frameworks
• Own compiler, runtime, and library strategy to map workloads to optimal compute paths
• Develop performance models for AI workloads to evaluate configurations before silicon
• Co‑design interface contracts, telemetry, and control for dynamic workload movement between compute paths
• Define scale‑up/scale‑out software strategies, including workload distribution across nodes and systems
• Prototype kernels, libraries, and runtime concepts to validate architectural decisions against AI benchmarks

• 5+ years of experience in compiler, runtime, or systems software for accelerated, heterogeneous, or specialized compute platforms
• Strong compiler architecture experience (LLVM, MLIR, or similar)
• Experience defining instruction sets or abstractions for novel compute architectures
• Background in kernel and library optimization for AI workloads across backends
• Experience modeling AI workload performance (training/inference) on different hardware
• Understanding of scale‑up/scale‑out systems software, scheduling, and accelerator control plane
• Ability to collaborate with hardware architects to define shared interfaces and system design decisions
• Experience with compiler/runtime/AI software ecosystems (LLVM/MLIR, Triton, PyTorch, JAX, Python, C++, profiling tools)
• Familiarity with photonic/optical compute concepts is a plus
• Experience with heterogeneous ecosystems (CUDA, ROCm, Triton, oneAPI, or custom stacks)
• Contributions to open compiler/runtime ecosystems or firmware‑level scheduling is a plus

• collaboration across cross-functional teams
• communication of complex concepts to engineers and stakeholders
• problem‑solving and analytical thinking
• LLVM
• MLIR
• Triton