Production / SRE / infrastructure engineer focused on reliable systems, Kubernetes/cloud operations, observability, and automation that is safe to run in production.
Upstream contributor to gVisor, KHI, go-containerregistry, syzkaller, Composer local dev, and Google SecOps MCP. My merged work covers Linux runtime behavior, Kubernetes diagnostics, container archive safety, and service lifecycle reliability.
| Project | Technical Contribution | Merged Evidence |
|---|---|---|
| gVisor | Honor OCI swap limits in pre-created cgroup v2 paths while preserving caller-owned limits; emit Linux-compatible netstat name/value rows. | #13276, #13275 |
| KHI | Detect truncated Kubernetes audit logs, invalidate stale manifest state, and preserve condition-timeline continuity with regression tests. | #697 |
| go-containerregistry | Harden flattened archive export against Windows-style traversal and drive-prefixed paths, with regression coverage for entries and relative links. | #2330 |
| Composer local dev | Fix Airflow dependency-file ownership under the runtime user; add configurable startup timeouts across the CLI and readiness checks. | #135, #136 |
| syzkaller | Parse Linux int3 crash reports, prevent syzbot reply loops, and surface build root causes in AI job errors. | #7376, #7420, #7388 |
| SecOps SOAR MCP | Guard cleanup after failed initialization and avoid creating HTTP sessions during shutdown. | #258, #260 |
| GKE samples | Mark NVIDIA DRA installer pods safe to evict so installer DaemonSets do not block autoscaler scale-down. | #2081 |
| Google Benchmark | Exclude errored repetitions from statistics and skip hardware-counter tests when counters are unavailable. | #2199, #2204 |
| Stellar Engine | Remove default-service-account project IAM grants in Terraform bootstrap projects. | #68 |
Full contribution record includes additional merged work in ADK Go, Vertex AI Creative Studio, and the OpenTelemetry Operator sample.
Live upstream searches: Google / Google Cloud Platform.
- Production context includes Meta monetization data infrastructure and SHEIN gateway infrastructure work.
- Experience around production gateways, Kubernetes/AKS-style platforms, Kafka, ZooKeeper, Elasticsearch, Terraform, runbooks, dashboards, and operational automation.
A runnable Kubernetes reliability lab for production-style workloads:
- OpenTelemetry trace collection and Kubernetes resource context
- incident replay for baseline traffic, cache-miss latency, dependency timeout, and rollout regression
- SLO-style reliability gate with published evidence reports
- GKE-shaped manifests for collector RBAC, PVC-backed queue storage, NetworkPolicy, PDBs, and sample workloads
- CI validation that regenerates evidence, checks JSON/YAML, runs tests, and enforces a Kubernetes production manifest review
An offline release go/no-go gate for production-adjacent rollouts:
- evaluates baseline and canary windows from committed rollout evidence
- checks error rate, p95 latency, burn rate, rollback events, pager alerts, change freeze state, and required approvals
- writes JSON for automation and Markdown for release records
- dependency-free Python package with unit tests and GitHub Actions CI
- Production changes that can be rolled out, observed, and rolled back.
- Automation with explicit inputs, validation, state, side effects, and retry boundaries.
- Reliability evidence: runbooks, dashboards, audit trails, tests, and incident reports.
- Practical open-source changes that reduce ambiguity for maintainers and users.
Python Go C++ Java SQL Bash Linux Kubernetes AKS GKE OpenTelemetry Terraform Ansible Nginx/APISIX Kafka ZooKeeper Elasticsearch CMake pkg-config GitHub Actions
- GitHub: Haihan-Jiang
- Engineering profile: haihan-jiang.github.io
- LinkedIn: haihan-jiang
- Email: haihanj99@gmail.com
Upstream merge status verified on 2026-09-07. Contribution evidence.