I'm a backend engineer at Alibaba (Qwen) in Guangzhou, working on web data infrastructure — large-scale page rendering and content acquisition services that sit between the open web and LLM applications.

Day to day I write Go, Java, Node.js/TypeScript and shell, and I care about reliability, observability, and keeping systems boring.

Contributing to Nacos, Redis and Prometheus. I enjoy the community side of open source and I'm interested in taking on more wherever I can be useful.

Nacos — server-side work and keeping the Node.js and Go SDKs consistent with the Java one:

• nacos server: log the resolved logging config location and add namespaceId to subscribe logs — alibaba/nacos#15792 (merged)
• nacos server: remove the prometheus dependency from nacos-client via a pluggable metrics SPI — alibaba/nacos#15793
• add selectOneHealthyInstance to the Node.js naming client — nacos-sdk-nodejs#152 (merged)
• align config local cache with the Java SDK (failover + snapshot lifecycle) — nacos-sdk-nodejs#154
• support publishConfigCas for optimistic-lock style config publish — nacos-sdk-nodejs#153 (merged)
• always subscribe services loaded from local disk cache — nacos-sdk-go#911 (merged)
• fix the default value of nacos.core.auth.enabled in the auth docs — nacos-group.github.io#1144
• add a Java SDK client metrics guide for the Micrometer migration (follow-up to alibaba/nacos#15793) — nacos-group.github.io#1147

Redis:

• fix keyspace notification event order for list move commands — redis/docs#3855 (merged)

Prometheus:

• alertmanager: make silence creator and comment optional — alertmanager#5471 (merged)
• procfs: model legacy TcpExt counters removed from recent kernels — procfs#863 (merged)
• node_exporter: use procfs parsers for netstat, snmp and snmp6 — node_exporter#3796 (merged)

Kubernetes — currently in review:

• kube-apiserver: use contextual logging — kubernetes#141778
• kube-controller-manager: use contextual logging in the clustertrustbundle publisher controller — kubernetes#141553
• batch: enable the commentstart kube-api-linter rule — kubernetes#141783
• networking: migrate IngressClass parameters scope to declarative validation — kubernetes#141682
• migrate Job non-negative spec fields to declarative validation — kubernetes#141680
• migrate Deployment and ReplicaSet selector to declarative validation — kubernetes#141558
• informer-gen: use plural-exceptions-aware namer for GVR resource name — kubernetes#141566

Other communities:

• better-harness: attribute verdict source and guard green runs in the project docs — better-harness#121 (merged)

Distributed systems, distributed configuration and service discovery, web infrastructure, developer tooling, AI applications, AI coding and evaluation harnesses, and open-source communities.