Principal Software Engineer - Accelerate DiagnosticsJanuary 2014 - September 2021
- Developed Clonetracker, the image processing component of the Pheno product. Clonetracker implements a traditional image processing pipeline to analyze bacteria in fluorescence and dark-field microscopy images in C++/CUDA. Working in a team of three, I was solely responsible for application performance and the production codebase from inception to 2018. It is in use in hundreds of hospitals.
- Designed, procured, and implemented the Opus Feasibility Computing Environment, an on-premises all-flash parallel filesystem, private high-speed ethernet network, and GPU-based batch processing system to support algorithm development for AXDX’s next AST instrument.
- Developed a software package for manual and automated interpretation of broth microdilution (BMD) plates. This software provided key reference data for the Pheno product. Automated interpretation based on optical density matched expert human performance.
- Lead statistician on the Pheno FDA De Novo clinical trial and submission. I designed the set of experiments, analyzed all data in R, provided results for the clinical study report, and negotiated labeling with FDA. I also led clinical trial design for the Pheno Bacterial Pneumonia (BP) test kit in which I successfully negotiated a 75% reduction in prospective samples required in the main arm of the study.
Senior Software Engineer - ArrayFireMay 2013 - December 2013
- Developed optimized CUDA kernels for clients’ production codebases (including MathWorks’s Matlab) and the ArrayFire open source library
Computer Scientist (Future Technologies Group) - Oak Ridge National LaboratoryJune 2009 - May 2013
Performed independent computer science research evaluating emerging computing technologies and supported DoE’s scientific applications. Select projects include:
- Aspen: A domain specific language for performance modeling designed to estimate the tradeoffs of large-scale scientific applications
- SHOC: A benchmark suite for GPU clusters including low-level performance primitives through kernels of scientific applications, implemented in CUDA, OpenCL, and MPI.
- Keeneland: NSF’s 5-year, Track 2D project to bring innovative computing architectures to the open science community. I was involved in all aspects of the system including acquisition, benchmarking, acceptance testing, and application tuning for the 264-node cluster.
- S3D: I developed a GPU-accelerated version of S3D, software for direct numerical simulation of turbulent combustion (Navier-Stokes) with complex chemistry, implemented in CUDA and MPI.
|M.S. Computer Science||Georgia Institute of Technology||2007-2009|
|B.S. Computer Science||Georgia Institute of Technology||2004-2007|
- Carbapenem vs. Piperacillin–tazobactam Definitive Therapy for Patients with Bloodstream Infections Due to Ceftriaxone Not Susceptible Escherichia coli or Klebsiella species. Open Forum Infectious Diseases, Volume 6, Issue Supplement 2, October 2019.
- Evaluation of Empiric β-Lactam Susceptibility Prediction among Enterobacteriaceae by Molecular β-Lactamase Gene Testing. Journal of Clinical Microbiology. September 2019.
- Keeneland: Computational Science Using Heterogeneous Computing. Book Chapter in Contemporary High Performance Computing: From Petascale Toward Exascale. CRC Computational Science Series. Taylor and Francis. 2017.
- Automated Design Space Exploration with Aspen. Scientific Programming. January 2015.
- Quantifying Architectural Requirements of Contemporary Extreme-Scale Scientific Applications. Proceedings of the International Workshop on Performance Modeling, Benchmarking and Simulation of High Performance Computer Systems. October 2014.
- Modeling Synthetic Aperture Radar Computation with Aspen. International Journal of High Performance Computing Applications. July 2013.
- Productive High Performance Software for OpenCL Devices. SPIE Modeling and Simulation for Defense Systems and Applications. 2013.
- Aspen: a Domain Specific Language for Performance Modeling. Proceedings of the International Conference on High Performance Computing, Networking, Storage and Analysis. November 2012.
- The Tradeoffs of Fused Memory Hierarchies in Heterogeneous Computing Architectures. Proceedings of the ACM International Conference on Computing Frontiers (CF12). May 2012.
- Quartile and Outlier Detection on Heterogeneous Clusters Using Distributed Radix Sort. Proceedings of the Workshop on Parallel Programming on Accelerator Clusters (PPAC). September 2011.
- Keeneland: Bringing Heterogeneous GPU Computing to the Computational Science Community. IEEE Computing in Science and Engineering. 13:5 2011.
- Non-Uniform Device Access Performance in Scalable Heterogeneous GPU Systems. IEEE MICRO, Special Issue on CPU, GPU, and Hybrid Computing. October 2011.
- Quantifying NUMA and Contention Effects in Multi-GPU Systems. Proceedings of the Fourth Workshop on General-Purpose Computation on Graphics Processors (GPGPU 2011). Newport Beach, CA, USA. March 2011.
- Maestro: Data Orchestration and Tuning for OpenCL Devices. Proceedings of the European Conference on Parallel Computing (EuroPar 2010). Naples, Italy. September 2010.
- The Scalable Heterogeneous Computing (SHOC) Benchmark Suite. Proceedings of the Third Workshop on General-Purpose Computation on Graphics Processors (GPGPU 2010). Pittsburgh, PA, USA. March 2010.
- Accelerating S3D: A GPGPU Case Study. Proceedings of the Seventh International Workshop on Algorithms, Models, and Tools for Parallel Computing on Heterogeneous Platforms (HeteroPar 2009). Delft, The Netherlands. August 2009.
- Accelerate Pheno System and PhenoTest BC Kit. Granted 2017-02-23. (DEN160032). First antibiotic susceptibility test from positive blood culture broth to receive FDA clearance.