AI Infrastructure · Edge Systems Engineering
Yuqing Ye
I build intelligent systems that have to run on real hardware — deploying vision models to onboard compute, wiring embedded devices to cloud inference, and shipping platforms people actually use.
- 70.2%
- Best mAP, onboard detector
- 14–15
- FPS on embedded target
- 100+
- Active users shipped to
Selected work
Three systems, from onboard perception to a campus-scale platform.
Edge AI Vision Deployment
Detect-and-Avoid perception on resource-constrained onboard computers.
A reproducible pipeline from training to onboard inference: quantization, benchmarking, and controlled RGB vs grayscale experiments.
- 70.2%Best mAP
- 14–15FPS onboard
- 0.82Precision
Intelligent Robotic Pet
An ESP32-powered companion joining embedded hardware, cloud intelligence, and interactive behavior in a palm-sized system.
Cloud-assisted architecture and streaming PCM audio, cutting perceived voice latency on hardware that cannot run an LLM locally.
- 6Subsystems integrated
- 1Persistent WebSocket link
- ~$30Bill of materials, prototype
UW-Social
A student-centered event discovery and community platform.
Solved a cold-start problem with deliberate scope narrowing and a newsletter-to-structured-data ingestion path.
- 300+Curated events
- 100+Active users
- 60+Discord members
About
Electrical & computer engineer working across edge AI, embedded systems, and product. I start from a constraint — latency, power, memory, cost — and design backwards.
Currently at ARC Lab deploying computer vision on resource-constrained drone platforms; previously built AI integrations for small businesses at Halogen.AI.