Page Not Found
Page not found. Your pixels are in another canvas.
Sitemap
A list of all the posts and pages found on the site. For you robots out there is an XML version available for digesting as well.
Pages
About me
Kailai Cui is a PhD researcher building resource-efficient multimodal AI systems through task-aware inference and serving.
Page not in menu
This is a page not in th emain menu
Posts
ACM SenSys 2022
Published:
From Nov 6 to 9th, I traveled to Boston to attend ACM SenSys where I shared my work with researchers and learned about the diverse research that others did. Some professors asked insightful questions about my project, for example, is an analytical model more suitable for the task? What knowledge does the machine learning model produce? From such conversations, I learned to place my research in the context of a broader applications and question its scientific value. I also worked as a student volunteer. This gives me the chance to talk to many students and know about their PhD life. 
2022 Fall Undergraduate Research Symposium
Published:
On September 30th, the 2022 Fall Undergraduate Research Symposium took place in Swem Library Read & Relay Room. Among more than 50 research posters presented by W&M undergraduates, I showcased our recent research project called “Light Auditor”. In this work, we are tackling IoT privacy via power side-channel auditing. A novel covert channel attack leaks user’s private data by encoding and transmitting them through smart bulb’s infrared emission. We used a power-auditing system and a CNN model that identifies the smart bulb’s leaking of private data.
portfolio
publications
Light Auditor: Power Measurement can tell Private Data Leakage through IoT Covert Channels
Published in ACM SenSys 2022, 2022
Light Auditor detects private-data leakage through IoT covert channels using power measurements and machine learning.
Recommended citation: Woosub Jung, Kailai Cui, Kenneth Koltermann, Junjie Wang, Chunsheng Xin, Gang Zhou, "Light Auditor: Power Measurement can tell Private Data Leakage through IoT Covert Channels," ACM SenSys, 2022. /files/nov2022paper1.pdf
UWB-Based Localization of Smartphones inside a Vehicle to Prevent Distracted Driving
Published in ACM/IEEE SenSys 2026, 2026
UWB-PTrac provides real-time, seat-level phone localization using existing UWB keyless-entry infrastructure with a single in-cabin, single-antenna anchor. It resolves spatial ambiguity through optimized anchor placement or a lightweight retrofit RF shield. City-driving tests achieved 96% driver-vs-other-seat classification accuracy with optimized anchor placement and 91% with the RF shield.
Recommended citation: Kailai Cui, Ke Sun, Kang G. Shin, "UWB-Based Localization of Smartphones inside a Vehicle to Prevent Distracted Driving," ACM/IEEE International Conference on Embedded Artificial Intelligence and Sensing Systems (SenSys), 2026. https://doi.org/10.1145/3774906.3800472
EventHOI: Event-Driven Energy-Efficient Hand-Object Interaction Logging on Smartglasses
Manuscript under review
EventHOI combines low-power event sensing, IMU-based egomotion cancellation, and interaction-aware compression to capture RGB evidence around hand-object interactions. Across 50 hours of data with more than 19,000 verified interactions, it reduced RGB frames by 50-65% and bandwidth by 2.7x compared with uniform 1-FPS RGB, with a 1.0-2.0 percentage-point accuracy loss and projected 33% daily power savings.
Recommended citation: Le Zhang, Kailai Cui, Hao Chen, Vlad Roznyatovskiy, Jianzhong Zhang, Anhong Guo, Kang G. Shin, Ke Sun, "EventHOI: Event-Driven Energy-Efficient Hand-Object Interaction Logging on Smartglasses." Manuscript under review.
Agentic Edge-Cloud Orchestration for Resource-Efficient Daily Food Inventory Monitoring with Egocentric Video
Manuscript under review
InvAgent treats daily food-inventory monitoring as sparse-evidence inference: brief observations reveal item use and remaining amount, while whole-video VLM analysis spends computation on irrelevant frames. An edge context builder summarizes item interactions, and a bounded agentic workflow selects amount-evidence windows and revisits unresolved items. Across approximately 39 hours from seven households, the system reduced VLM input tokens by 9.9x versus 1-FPS whole-video inference while increasing amount-aware inventory F1 from 66.6% to 68.9%.
Recommended citation: Kailai Cui, Kaylee Yaxuan Li, Chaoyu Zhang, Peizhou Huang, Hao Chen, Jianzhong Zhang, Ke Sun, Kang G. Shin, "Agentic Edge-Cloud Orchestration for Resource-Efficient Daily Food Inventory Monitoring with Egocentric Video." Manuscript under review.