Job SummaryThe AI/ML ASIC Architect will lead the design and optimization of advanced ASIC and SoC architectures to support large-scale machine learning workloads. This role requires deep expertise in I/O subsystems, accelerators, memory hierarchies, and AI/ML model optimization. The architect will collaborate with cross-functional teams to deliver scalable, high-performance solutions that meet airworthiness, reliability, and enterprise requirements.
Key Responsibilities- Design ASIC/SoC architectures including PCIe, UCIe, CXL, and DMA engines.
- Integrate I/O subsystems with embedded processors (x86, RISC-V, ARM).
- Architect GPU/TPU/xPU accelerators with high-bandwidth memory hierarchies.
- Optimize LLM training/inference including Dense, Mixture of Experts (MoE), and multi-modal architectures.
- Implement KV cache optimization, Flash Attention, and MoE techniques.
- Drive ML system optimization for large-scale GPU and ML architectures.
- Apply microarchitecture principles to enhance performance engineering.
- Collaborate on compliance with airworthiness and enterprise standards.
Required Qualifications- Bachelor's degree in Mechanical, Aerospace, or Computer Engineering (BSME, BSAE, BSCE).
- Minimum 9 years of relevant experience; OR Master's degree with 7 years of experience.
- Expertise in ASIC, SoC, and I/O subsystem architecture.
- Strong knowledge of ARM processors and AXI interconnects.
- Experience with GPU/TPU/xPU accelerators and ML system optimization.
- Understanding of LLM architectures, KV cache optimization, and Flash Attention.
- Proficiency in microarchitecture principles for performance engineering.
- Preferred Qualifications
- Experience with UCIe, CXL, NVLink, or UALink protocols.
- Knowledge of high-speed networking (InfiniBand, RDMA, NVLink).
- Familiarity with transformer architectures, attention mechanisms, and model parallelism.
- Hands-on experience with CUDA programming and GPU memory hierarchies.
- Exposure to large-scale distributed training systems.
- Experience with NVMe storage systems, NAND flash, and firmware/ASIC design.
- Proficiency in Matlab or Mathcad for analysis.
- Strong communication and leadership skills.
Certifications- None required; certifications in ASIC design, ML systems, or GPU programming are a plus.
