Kory Watson
I'm an engineer specializing in AI / ML performance, compilers, optimization, and quantization.
I co-created and maintain AI Hub Models and AI Hub Apps - open-source projects providing 175+ optimized ML models and sample applications for AI deployment on the edge.
For inquiries or to otherwise get in contact, shoot me an email at [email protected]. Thanks!
Experience
Qualcomm Staff Machine Learning Engineer 2023 - Current
- Co-created AI Hub Models and AI Hub Apps, the largest collection of edge-deployable models on Huggingface with fully open source recipes on GitHub
- Managed model contributions from 16+ engineers; developers can optimize, validate, and deploy AI models on Qualcomm-powered devices within minutes
- Built an ML benchmarking suite that drove improvements in graph compilation coverage, latency, and accuracy across all Qualcomm software and hardware
- Drove performance improvements by identifying and resolving root causes of performance regressions, resulting in significant latency reductions across many graphs
Tetra AI (Acquired by Qualcomm) Machine Learning Engineer 2022 - 2023
- Member of founding team focused on Neural Network Optimization for Mobile and Edge
- Made core contributions to the Qualcomm AI Hub edge neural network compiler
- Designed core services and developer-facing APIs that power the Qualcomm AI Hub user experience today
- Worked with customers including Microsoft, Verizon, Roblox, and Sonos on latency-sensitive real-time video, image, and speech applications
- Helped customers deploy models with 2-10x improvements in latency, battery life, and memory in real workloads
Apple Machine Learning / Compiler Engineer 2020 - 2022
- Built a compiler and runtime for blazing-fast model inference and training on the edge
- Core contributor to MIL, an intermediate representation similar to MLIR for ML models; designed a general ML opset
- Designed & implemented a language serializer / parser to translate to/from a readable C-like format
- Enhanced lowering code for hardware compiler IRs, implementing optimizations for significant performance gains
- Helped design software that breaks a model into pieces that run on optimal hardware backends
Education
University of WashingtonComputer Engineering2016 - 2019
Internship Experience
| Apple | Applied ML Research | Machine Learning Engineer Intern, Vision | 2019 |
| Apple | Applied ML Research | Software / Machine Learning Engineer Intern, Vision | 2018 |
| Cloud Billing | Software Engineer Intern | 2017 | |
| Microsoft | SQL Server | Software Engineer Intern | 2016 |
| Blue Origin | Avionics | Software Engineer Intern | 2015 |
| Boeing | P8 Program | Software Engineer Intern | 2014 |
| SeaTec Consulting | Airborne Wifi | Lead Software Engineer Intern | 2014 |
Teaching (TA)
| CSE 401 | Intro to Compiler Construction | Abstract Syntax Trees, Finite State Machines, Grammars, x86 Assembly, IRs, etc. | Fall 2019 |
| CSE 333 | Systems Programming | C and C++ Intro, Berkeley Networking API | Spring 2019 |
| CSE 351 | The Hardware/Software Interface | x86 Assembly, Computer Architecture, Caching, Virtual Memory, Memory Allocation | Autumn 2018, Winter 2019 |