Principal Software Architect- High Performance Computing
Core
Design and implement robust, scalable HPC infrastructure solutions combining CPUs, GPUs, and FPGAs for semiconductor equipment image processing and deep-learning workloads.
Role type
Principal Software Architect (High Performance Computing)
Builds
High-performance computing solutions for low-latency image processing and AI inference on chip manufacturing process control equipment.
Domain
Semiconductor manufacturing / High Performance Computing / Heterogeneous Computing
Deliverable
production ML models
Required skills
HPC system design, CUDA programming, parallel programming (MPI, OpenMP, OpenACC), GPU inference server management (Nvidia Triton), distributed computing, performance profiling (vtune, perf, Nsight), low-level code optimization, microservices architecture, containerization (Docker/Singularity), RDMA/RoCE/InfiniBand, cluster management (SLURM)
Preferred skills
Vectorization, Intrinsics, cache-aware programming, lock-free data structures, DAG execution frameworks (Intel TBB)
Technologies
CUDA, MPI, UCX, Nvidia Triton, Docker, Singularity, SLURM, InfiniBand, RDMA, OpenMP, OpenACC, OpenCL, C/C++
Responsibilities
Analyze and partition workloads to appropriate compute units (CPU vs GPU/FPGA); prototype designs with real code and data; profile clusters and nodes to identify bottlenecks; conduct performance tuning and capacity planning; evaluate and recommend technologies; mentor junior engineers.
Seniority
Principal, hands-on IC with mentorship
