Alex Shroyer
PhD in machine learning + 10 years in embedded systems, ML, DevOps, and education
Professional Experience
Cato Digital
Senior DevOps Engineer/SRE November 2024 -- present
Building obtainable low-carbon cloud compute/storage/GPU infrastructure.
- Backend devops, running Elixir and Postgres in Docker containers
- Salt change management, Wedge switch configuration
- Quality of life scripts in Python and Bash for automating repetitive tasks
- Implement live alerting with self-hosted Grafana
- NVIDIA DGX-2 bare-metal deployments and Open Compute Project hardware
- Codify policies into code when possible, but reinforce with a culture of documenting and sharing knowledge
Indiana University Intelligent Systems Engineering
for teaching experience, see: teaching
Staff Systems Engineer February 2021 -- November 2024
Developed Python-based deep learning models in PyTorch and Tensorflow and created datasets using diverse compilation with Docker and QEMU for obfuscated compiled binary code research and anti-malware analysis.
- Created binary analysis dataset of 30k+ embedded code samples using Ghidra, Tigress, and angr
- Advanced obfuscated binary code analysis by 4x using basic block embedding vectors
- Customized and administered Docker and Django-based SaaS, growing to thousands of users
- Built network sandbox for testing/throttling TCP/IP and UDP client-server applications over IPv4
Senior Electronics Engineer July 2016 -- February 2021
Built datacenter cluster, server, power, and switch hardware and supported hardware and software infrastructure for Engineering courses to run and test untrusted code safely in a sandboxed environment.
- Built Xilinx FPGA and Nvidia GPU clusters (from 8 to 64 nodes) for high performance network and graph analytics
- Simulated network faults (latency and packet loss) for 200+ student projects concurrently
- Developed hardware-accelerated distributed edge computing prototypes and publications
- Taught engineering courses (Computer Architecture, C, Python, Operating Systems, Networks)
Indiana University Psychological and Brain Sciences
Electronics Engineer March 2014 -- June 2016
Built custom hardware and software for embedded devices used in brain science research.
- Created novel MRI-safe touch screen technology (awarded US Patent 10820839B2)
- Developed sensor systems for research: skin response, eye tracking systems, brain-computer interfaces
- Worked directly with customers through full product life cycle from concept to production
- Built prototyping labs including 3D printer and laser cutter operations and training
WSO2, LLC
Software Engineer January 2014 -- March 2014
Supported enterprise middleware applications based on Apache products.
- Used SOAP and REST frameworks
- Learned Java-based open source frameworks
Education
Ph.D. in Intelligent Systems Engineering
Indiana University (2018-08-20 – 2023-12-15). Computer Engineering concentration, Computer Science minor. Dissertation: Deep Learning for Obfuscated Code Analysis.
Bachelor of Science in Electrical Engineering
Indiana University Purdue University Indianapolis (2010-05-11 – 2013-12-16).
Skills
- (language) Python, Go, C/C++, CUDA, JavaScript, Bash, Racket, Elm, Scala, K, Q
- (framework) PyTorch, Tensorflow, FastText, Gensim, OpenCV, Scikit-Learn, React
- (platform) Docker, Google Cloud Platform (GCP), Matplotlib, Seaborn, Linux, Google Apps Script
- (utility) PostgreSQL, Grafana, SQLite, Salt, Django, Pandas, Git, Unix socket programming, Vite
- (conceptual) project management, technical writing, teaching, vector embedding, NLP, AI, ML
Projects
- ObfuscatedBinaryClassifiers: dissertation datasets plus Python and C source code
- Element: experimental array language interpreter for GPUs using Python and CUDA
- Open source contributions to PyTorch and Tensorflow
- HDL Online: easily edit and share Nand2Tetris HDL code with syntax highlighting
- Prototype GPU support for the J programming language using ArrayFire library (paper and code)
Publications
- Shroyer, Alexander, and D. Martin Swany. "Detecting Standard Library Functions in Obfuscated Code." IntelliSys, September 2023. Best presentation award (video).
- Shroyer, Alexander, and D. Martin Swany. "Data Augmentation for Code Analysis." 2022 International Conference on Intelligent Data Science Technologies and Applications (IDSTA). IEEE, 2022.
- Brasilino, Lucas RB, Naveen Marri, Alexander Shroyer, Catherine Pilachowski, Ezra Kissel, and Martin Swany. "In-network processing for edge computing with InLocus." International Journal of Cloud Computing 9, no. 1 (2020): 55-74.
- Brasilino, Lucas RB, Alexander Shroyer, Naveen Marri, Saurabh Agrawal, Catherine Pilachowski, Ezra Kissel, and Martin Swany. "Data Distillation at the Network's Edge: Exposing Programmable Logic with InLocus." In 2018 IEEE International Conference on Edge Computing (EDGE), pp. 25-32. IEEE, 2018.
- Arap, Omer, Lucas RB Brasilino, Ezra Kissel, Alexander Shroyer, and Martin Swany. "Offloading collective operations to programmable logic." IEEE Micro 37, no. 5 (2017): 52-60.
- "Thinking in Array Languages" with Richard Feldman on Software Unscripted, July 8 2023.