Anomaly Detection and Log Analysis, Spring 2019
Worked as an research intern at Beijing Bizseer Technologies to develop Deep learning based Anomaly Detection and Log Analysis Algorithms to help major commercial banks in china find out Anomaly in daily data and uncover root cause in massive heterogeneous data and log reports of their complex and critical systems IT infrastructure.
Azure Capacity Management, Spring 2018
Developed solutions to predict capacity days-to-exhaustion and Develop Buffer Capacity Management strategy to reserve the empty node/core buffer to fit the tenant upgrade needs. Work on feature engineering and data pre-processing. Investigate AutoML's performance and compare the result with tuned models by domain experts.
Azure Node Failure Prediction, 2017
Co-authored a Paper: Predicting Node Failure in Cloud Service Systems at ESEC/FSE 2018.
A cloud service system typically contains a large number of computing nodes. In reality, nodes may fail and affect service availability. In this paper, we propose a failure prediction technique, which can predict the failure-proneness of a node in a cloud service system based on historical data, before node failure actually happens. The ability to predict faulty nodes enables the allocation and migration of virtual machines to the healthy nodes, therefore improving service availability.
- Work on Data Analytcis, and developed an effective ensemble model for production.
- Developed a production-ready predictive service for engineering teams.
- Achieved 43% reduction on Azure VMs’ failure rate.
BoneyCare is designed to help people with speech disorders improve their conﬁdence and speaking quality. The app records, transcribes and analyzes a users speech patterns, which speech pathologists can then use to improve and accelerate their diagnoses and treatment plans. This project is a collaboration between 2nd year computer science students from two Chinese universities and UC Berkeley.
First app targeting stutter with cloud-based machine learning, voice recognition, wave analysis, and emotion recognition technology.
Provide a new way to help speech pathologists connect and track their patients status.
Ability Award Winner at Imagine Cup World Final.
LightItUp is a hackathon solution based on MITM(man-in-the-middle) attack on DL/T645-2007 protocol that helps late-owl students gain access over their dorm's DL/T645-2007 compliant electricity meter and bypass enforcement of campus electicity after 23:00.
Sucessfully gained remote control of campus electricity command and control device.
Conducted Proof Of Concept Attack and controlled 30s of black out across the whole dorm building.