Ye Wang

Designed and trained unsupervised Byzantine fault tolerant clustering algorithm to detect malicious nodes in global network with 30M+ nodes, achieving 95% accuracy. Integrate the algorithm to the verification layer of a Proof of Location protocol.
Engineered MySQL/Python pipelines processing 300M/day geo-based ping records, and built an internal dashboard to monitor global connection network and latency distribution.
Designed a reward model with sigmoid-curve optimization to boost honest nodes’ participation.