Solution Architect — Core ML Frameworks, Kernels & Performance

About the role

We are looking for a Solution Architect to lead the design and optimisation of high‑performance kernels for Core ML frameworks. The role focuses on accelerating both training and inference on the latest TPU and GPU hardware, directly impacting Google Cloud AI services and the open‑source ecosystem.

Key responsibilities

• Design and optimise kernels using languages such as Pallas, Mosaic and Triton for TPU and GPU architectures.
• Build and maintain benchmarking suites, autotuning frameworks, performance analysis tools and regression tests.
• Stay up‑to‑date with hardware, compiler and AI model advances to identify optimisation opportunities.
• Collaborate with ML researchers, framework developers (JAX, PyTorch) and compiler engineers (XLA) to drive adoption and resolve bottlenecks.
• Produce clear documentation and developer‑facing APIs for open‑source libraries.

Required profile

• 12+ years of industry experience, with at least 5 years in C++ or Python development.
• 3+ years of experience testing, maintaining or launching software products and 1+ year in software design/architecture.
• Proven track record of building developer infrastructure and open‑source libraries.
• Strong investigative, problem‑solving and communication skills across cross‑functional teams.

Required skills

• C++
• Python
• Pallas
• Mosaic
• Triton
• CUDA
• TPU
• GPU
• JAX
• PyTorch
• XLA
• MLIR
• OpenXLA
• Kernel‑level performance optimisation
• Understanding of modern accelerators and compiler principles

What we offer

• Opportunity to work on cutting‑edge AI hardware and software at scale.
• Collaboration with world‑leading researchers and engineers.
• Remote work option for candidates based in Slovakia.