Summary
Yunseong Lee is a Senior Staff Engineer with 11 years of experience specializing in ML/DL inference systems and large-scale data processing, currently leading engineering efforts at Qualcomm. He holds a Ph.D. candidacy from Seoul National University and has a strong research-to-product track record building systems like Pretzel (white-box ML inference), RnB (multi-GPU DNN pipeline replication and batching), and Splash (collaborative inference for edge devices). His work blends systems and model-aware optimizations—exploiting training-time knowledge to optimize end-to-end execution and using compression and hardware-specific acceleration (e.g., DSP/NPU) to run heavy models on weak devices. Yunseong’s background includes internships at Microsoft and Microsoft Research Asia and a continuity of advancing roles at Qualcomm, reflecting both deep research roots and production-grade engineering impact. An understated strength is his focus on heterogeneous-resource scheduling across CPUs, GPUs, and embedded accelerators, which consistently improves latency and throughput in deployed inference workloads.
11 years of coding experience
12 years of employment as a software developer
PhD Computer Science and Engineering, PhD Computer Science and Engineering at Seoul National University