# Seungil Lee

> Software Engineer

Location: Walnut Creek, California, United States
Profile: https://flows.cv/seungil

Senior Software Engineer with 20 years of expertise in high-performance C++ and system-level optimization. Proven track record in developing and tuning ML runtimes on mobile devices, with extensive experience in ONNX Runtime, TFLite, and PyTorch. Specialist in maximizing hardware utilization and minimizing memory footprint through low-level performance tuning, ensuring deterministic performance without relying on heavy abstractions. Expert in navigating resource-constrained environments, from developing high-efficiency execution engines to optimizing latency-critical system software.

## Work Experience
### Software Engineer @ Zoox
Jan 2024 – Present | Foster City, CA
· Offline Regeneration System Optimization: Enhancing a high-performance C++ toolset used
to re-execute on-vehicle sensor data and system states in offline environments for debugging
and performance analysis.
· Performance Tuning & Scalability: Identifying and resolving critical system bottlenecks to
improve data throughput and execution speed, directly accelerating the development iteration
cycles for autonomous driving software.
· Infrastructure Management: Streamlining the integration and deployment of regeneration
tools by maintaining a robust Bazel-based build system

### Software Engineer @ Grammarly
Jan 2018 – Jan 2024 | San Francisco Bay Area
· Cross-Platform ML Inference Library: Designed and developed a unified C++ inference library
for iOS and Android, integrating ONNX Runtime to execute complex NLP models on mobile
devices.
· Full-Stack Runtime Implementation: Developed high-performance C++ modules for
pre-processing and post-processing, ensuring consistent model execution across different
mobile operating systems.
· Native API Bridge & Wrappers: Architected and implemented Swift (iOS) and Java/JNI
(Android) wrappers to expose core C++ library functionalities, providing seamless integration
for native mobile applications.
· CPU-Centric Performance Tuning: Optimized inference latency and memory footprint
through low-level C++ tuning and efficient memory management, tailored for CPU-based
execution under strict mobile resource constraints.

### Senior / Principal Software Engineer @ UBIVELOX
Jan 2000 – Jan 2018
· JavaCard VM Development: Architected and implemented a high-performance JavaCard
Virtual Machine (VM) for secure elements, focusing on robust execution within extremely
limited memory (RAM/Flash) and strict security constraints.
· Mobile Java Virtual Machine (JVM): Developed and optimized mobile JVMs for feature phone
platforms, ensuring efficient byte-code execution and system stability across diverse hardware
specifications.
· Low-level Optimization & Assembly: Mastered C and Assembly (ARM) to optimize
performance-critical sections of the VM and system libraries, maximizing execution speed on
resource-constrained microcontrollers.
· Embedded System Architecture: Led the design of embedded OS components and runtime
environments, managing the full development lifecycle from architectural design to mass
production and deployment.


## Education
### Ph.D. in Computer Engineering
Seoul National University

### Master's degree in Computer Engineering
Seoul National University


## Contact & Social
- LinkedIn: https://linkedin.com/in/seungillee

---
Source: https://flows.cv/seungil
JSON Resume: https://flows.cv/seungil/resume.json
Last updated: 2026-04-11