> Curriculum Vitae
v1.2.5 - Q4 2025 | Download PDF
Education
Doctor of Philosophy in Computer Science
- Dissertation Title: Hardware Awareness for the Selection of Optimal Iterative Linear Solvers
Master of Science in Computer Science
Bachelor of Science in Computer Science, minor in Mathematics
Work & Research Experience
AI Performance Engineer, TPU Inference @ Google
- Architected and implemented Paged Attention for MaxText from design to merge, resolving complex concurrency bugs and achieving a 47% throughput increase in microbenchmarks.
- Developed custom Pallas kernels for 2D block-wise and sub-channel quantization, enabling advanced precision support within the TPU-Inference framework.
- Built model accuracy benchmarking framework for tracking long-term improvements and regressions of TPU-Inference models.
Machine Learning Engineer, Web-based QA @ Amazon (Alexa AI)
- Reduced overall latency of our model by 6x compared to the model’s baseline on GPUs.
- Responsible for performance analysis and optimization of a transformer-based research DL model.
- Created codebase compiling our model to ONNX, TensorRT, and Inferentia formats and ran associated benchmarking.
Research Engineer @ Amazon Web Services (AWS HPC)
- Created automated performance regression testing system for AWS HPC infrastructure.
- Benchmarked prototype EC2 instances to identify optimal hardware configurations for future HPC endpoints.
Research Engineer, ASR @ Amazon Alexa
- Created a new team to maintain an in-house deep learning framework and address organizational needs.
- Optimized distributed programming workflows using C++, Python, CUDA, TensorFlow, and MxNet.
Software Development Engineer, Mobile Hub @ Amazon Web Services
- Built Java-based backend services and data collection pipelines for customer usage analytics.
Doctoral Research Assistant @ University of Colorado, Lighthouse Project
- Utilized runtime performance data from supercomputers to train ML algorithms that predict optimal iterative linear solvers (C++, Trilinos, Python).
Computation Intern @ Lawrence Livermore National Laboratory
- Optimized BLAST hydrodynamics code for future architectures and developed benchmarking suites for high-performance linear algebra libraries using HPCToolkit.
Software Developer/Researcher @ TerraSpark Geosciences
- Implemented GPU-based seismic interpretation solutions using OpenCL, reducing processing time from hours to seconds compared to original CPU code.
Skills
- Languages: Python, C++
- Libraries & Frameworks: JAX, Pallas, CUDA, MPI, OpenMP, OpenCL
- Tools: XProf, DDT, ARM MAP, Intel Vtune, HPC Toolkit
Patents & Selected Publications
- US-12093669-B1: Massively parallel compilation of application code (Issued 2024-09-17)
- Publication: E. Jessup, P. Motter, et al., "Performance-Based Numerical Solver Selection in the Lighthouse Framework," SIAM Journal on Scientific Computing, 2016.
- Publication: K. H. Koh... P. Motter, "Will it stick? exploring the sustainability of computational thinking education," SIGCSE 2013.