Senior Compiler Engineer
Innatera Nanosystems2 months ago
Remote, United Kingdom or Remote, IrelandSenior
Responsibilities
- Design and implement compilers for next-generation AI accelerator architectures.
- Enable new neural-network topologies while optimizing chip performance and energy efficiency.
- Collaborate with SDK, algorithm, architecture, and hardware teams to define compiler requirements.
- Provide feedback to hardware engineers to improve future architectural designs.
- Research neural-network compiler technologies and introduce innovative solutions and industry best practices.
Requirements
- Proficiency in Python and C++ programming.
- Knowledge of neural networks and standard model architectures.
- Hands-on experience with PyTorch or TensorFlow.
- Hands-on experience with neural-network compiler frameworks such as MLIR or TVM.
- Strong understanding of computer architecture and practical experience optimizing performance for targeted hardware.
- Ability to rapidly prototype and test new ideas.
- Strong interpersonal and communication skills.
- Preferred: experience with IREE, ExecuTorch, ONNXRuntime, or compiler backend code generation using GCC or LLVM.
- Preferred: experience with RISC-V, especially its vector extension.
- Preferred: knowledge of digital or analog electrical circuits, spiking neural networks, embedded platforms and runtimes, graph algorithms, constraint solvers, or heuristic algorithms.
Benefits
- Innatera is an equal opportunity employer committed to a diverse, inclusive, and respectful workplace.
- Recruitment-process accommodations are available upon request.
About Innatera Nanosystems
Innatera Nanosystems designs and sells ultra-low-power neuromorphic processors and software for sensor-rich devices at the edge, targeting IoT, wearable, industrial, and automotive applications. Its brain-inspired analog-mixed-signal chips perform real-time inference close to sensors to cut latency and energy use for always-on analytics. The company is privately held, headquartered in Rijswijk, South Holland, and was founded in 2018 as a spin-off from Delft University of Technology.