Stateful Scheduler TeamOne of the founding engineers. Built the stateful Kubernetes operators and a customized kube-scheduler for LinkedIn's stateful apps like Kafka, Zookeeper, Espresso, Liquid, and other data systems, integrating Kubernetes with Helix, LinkedIn shard management system. My responsibilities included system design, protocol development, mentoring team members, and ensuring smooth integration with existing infrastructure. This initiative significantly improved LinkedIn's data handling capabilities and is LinkedIn next generation of Infrastructure.Compute Infrastructure team- Worked in a compute infrastructure team, dedicated to providing top-tier compute resource allocation and application deployment services at LinkedIn (LinkedIn Version Kubernetes Framework).- Spearheading LinkedIn's Platform as a Service, catering to approximately 8000 engineers.- Specializing in resource allocation and instance autoscaling for optimal performance.- Managing a large-scale fleet, ensuring efficient operation and maintenance.- Developing robust deployment infrastructure, encompassing both control and data planes.- Focusing on deployment reliability and implementing solutions for unhealthy host remediation.

- Worked for Kubernetes team about migrating all stateless app and services in LinkedIn platform from internal framework to Kubernetes to increase resource utilization by 30%.- Applied Admission Webhook to intercept Kubernetes Custom Resource Definition(CRD) requests to the API server and provided automation and safeguard on LinkedIn Kubernetes platform covering 25% users’ HTTP requests.- Designed and implemented the schema validator to check the requests by using Kubebuilder to prevent system crash.- Participated team on-call issues and helped users to run their product service on Kubernetes clusters.

- Worked on VMware Horizon about a component of lifecycle management for VMs in different cloud environments.- Used Java and RxJava to implement Inventory updates in the situation of restarting a powered-off VM on ZeroCloud/Azure.- Developed 50 unit tests and changed 5000+ lines of code to improve code coverage to 70% based on actions to VM by utilizing Cucumber and Junit.- Collected and cleaned 200,000 CPU and memory usage records from cluster through Wavefront and Swagger and helped to plan when/where to run chaos experiments.

- Worked for middleware team in OPPO Cloud about a store system to replace Redis and its auxiliary control platform.- Used Swagger to automatically generate API documents and avoided manual maintenance on API documents.- Designed new Grafana monitoring pages and metrics through PromQL and developed automatic registration function and simplified 90% of manual operations.- Conducted a performance test of the distributed storage system in Intel non-volatile memory and SSD and found a 30% efficiency improvement in read and write rates.

- Automatically acquired historical data on football matches by using Xpath and Python Request.- Utilized Python Selenium to enable automatic login and user status change and abnormal situations handling.- Analyzed football lottery impact factors and stored 330,000 records in MySQL database.