Summary
Konstantinos Parasyris is a Senior Compiler Architect with 12 years of experience building LLVM/Clang/MLIR-based compiler and runtime systems for GPU-accelerated HPC, currently at Intel after a multi-year research leadership role at Lawrence Livermore National Laboratory. He designs production-grade tooling—JIT specialization, kernel record–replay, and autotuning systems—that make performance, optimization decisions, and correctness observable and reproducible at scale. His work spans CUDA, HIP, and OpenMP offload portability and includes integrating ML-driven surrogate models into scientific workflows to accelerate tuning and inference. Earlier research focused on energy-efficient and significance-aware computing, and he helped create a CUDA–OpenMP cross-vendor offload framework and GPU-aware checkpoint–restart for resilience. Known for bridging research prototypes and production-quality compilers, he is an active contributor and reviewer within the LLVM and OpenMP ecosystems. Based in Livermore, CA, he combines deep compiler analysis with practical runtime engineering to push scalable HPC systems forward.
12 years of coding experience
12 years of employment as a software developer
Master's degree, Department of Electrical and Computer Engineering, Master's degree, Department of Electrical and Computer Engineering at University of Thessaly