Petr Petrenko
Engineering Manager & Platform Architect
Vice Manager, Platform SRE Group · Rakuten Group, Inc.
Vice Manager, Platform SRE Group · Rakuten Group, Inc.
·
Profile
Engineering leader and platform architect with 10+ years building cloud-native systems at scale. Currently managing two platform teams at Rakuten (Point & Coupon) handling peaks of up to 80,000 RPS, while continuing hands-on architecture work across GCP, Kubernetes, and distributed systems. Strong believer in autonomous systems over automation — building platforms that scale, heal, and operate themselves without human intervention. Built an AI-powered monitoring system on Gemini that runs fully autonomously across the entire infrastructure stack. Specialized in Kubernetes, IaC, observability, and turning "it should work" into reality under production pressure. Author of spanner-etcd — open source drop-in etcd replacement backed by Google Cloud Spanner (15× write throughput, 24h K8s soak test). Contributor to KEDA (GCP Spanner scaler).
AI Projects
Autonomous Infrastructure Monitoring System
2025 – Present
Rakuten Group, Inc.
- Designed and built a fully autonomous monitoring system powered by Gemini, covering the complete infrastructure stack: service latency and error rates, container CPU/memory, Kubernetes nodes, etcd, Kafka, Spark, Cloud Spanner, CloudSQL, SSL certificates, BigQuery, VPA recommendations, and pod colocation.
- Produces structured 24h ops summaries with severity-based triage (Needs Attention / Watch / All Clear / Suggestions) — correlating signals across all layers without human intervention, reducing alert fatigue and accelerating incident response.
- Operates fully autonomously in production for Japan Post platform (Yucho Points) — zero manual configuration required after deployment.
Open Source
spanner-etcd
2026
Author · Apache 2.0 · Go
Drop-in etcd v3 replacement backed by Google Cloud Spanner
- PENDING_COMMIT_TIMESTAMP revision — eliminates the global write lock, 15× throughput at ×32 concurrency
- Spanner Change Streams for Watch — ~30ms end-to-end latency
- Covering index optimization — +167% mixed workload performance
- Validated: Kubernetes v1.33.12, 24h soak test, zero crashes, zero data loss
- Directed reads support for multi-region Spanner deployments (
--spanner-read-location)
KEDA
2026
Contributor · ⭐ 16K+
GCP Cloud Spanner scaler
- New
gcp-spannertrigger — scales Kubernetes workloads from Spanner SQL query results - Supports Workload Identity Federation, service account JSON, env-based credentials
- 14 unit tests + 6 integration tests (emulator) + e2e tests
Skills
Orchestration
Cloud Platforms
Google Cloud
IaC & CI/CD
Observability
Networking
Data & Messaging
AI & Automation
Programming
Experience
Rakuten Group, Inc.
Feb 2026 – Present
Vice Manager, Platform SRE Group
Tokyo, Japan
- Managing two platform engineering teams (~20 engineers each) — Rakuten Point (peak 80,000 RPS) and Rakuten Coupon (peak 60,000 RPS) — overseeing day-to-day operations, release quality, and team performance.
- Driving engineering excellence: establishing release processes, conducting code and architecture reviews, and raising the technical bar across both teams.
- Continuing hands-on architecture work in parallel with management — currently leading architecture of a Points & Campaign platform for Telkomsel (Indonesia's largest telco, 169M subscribers), the largest system of this kind built at Rakuten.
- Leading hiring: conducting technical interviews, defining evaluation criteria, and growing the team with strong engineering talent.
- Designed fully autonomous scaling across the stack — GKE Autopilot, KEDA (traffic-driven), and Cloud Spanner autoscaling — eliminating the need for capacity planning or manual intervention entirely.
- Leading blameless postmortem culture across both teams, driving systemic improvements over individual blame.
- Resolving complex production incidents hands-on; providing Tier-1 on-call support with zero customer impact across multiple platforms.
Rakuten Group, Inc.
Oct 2023 – Feb 2026
Senior Technology Architect
Tokyo, Japan
- Architected an autonomous Point System for Japan Post (Yucho Points, millions of users) on GCP from scratch — GKE MultiCluster Gateway, Cloud SQL, Spanner, BigQuery, Dataflow — fully Terraform-managed; self-operating with minimal maintenance.
- Designed a Business Intelligence pipeline (Apache Astra → Cloud Spanner → BigQuery) federated for downstream AWS Glue analysis.
- Implemented Canary Releases across multiple platforms using Argo Rollouts with Kayenta-style analysis, gradual traffic shifting via Gateway API/Istio, and automated rollbacks.
- Introduced event-driven autoscaling across multiple platforms, cutting infrastructure costs by over 50% with no performance impact.
Rakuten Group, Inc.
May 2022 – Oct 2023
Solutions Architect & Architect Team Lead
Dalian, China → Tokyo, Japan
- Led live traffic migration from legacy HAProxy (with complex Lua-based routing logic) to the new One Point Platform — executed under full production load with zero downtime.
- Implemented AKS cluster autoscaling and QPS-based HPA, minimizing standby costs while enabling rapid failover.
- Built a traffic mirroring framework to validate production traffic in pre-production before go-live.
- Implemented Istio-based rate limiting to protect legacy systems under high concurrency.
- Delivered full-stack observability using OpenTelemetry, Jaeger, and Elasticsearch across Istio, APIs, workers, and batch systems.
- Diagnosed and resolved critical production issues: Cassandra LWT bottlenecks, AKS performance degradation (2× QPS gain), Tombstone overloads via automated compaction pipelines.
- Built a smart Kubernetes sidecar controller dynamically adjusting CronJob schedules based on Kafka lag and system dependencies.
- Developed incident-driven automation for self-healing failed transactions and automatic impact reporting.
Rakuten Group, Inc.
Mar 2021 – May 2022
Associate DevOps Architect
Dalian, China
- Built and managed a hybrid Kubernetes cluster for CI/CD, monitoring, artifact storage, and container registry across on-prem and cloud.
- Designed a hybrid monitoring platform (AKS + on-premises) using Prometheus and Thanos — adopted company-wide as the standard observability solution.
- Developed fully automated DevSecOps pipelines integrating Vault, Jenkins, ArgoCD, SonarQube, Harbor, and Trivy.
- Implemented IaC for Rakuten Points using Terraform, Ansible, and Python for on-demand provisioning and teardown.
- Deployed a Chaos Engineering platform for resilience testing across multiple Kubernetes clusters.
Rakuten Group, Inc.
May 2020 – Mar 2021
SRE / DevOps Engineer
Dalian, China
- Built a hybrid Kubernetes cluster spanning OneCloud and Azure for the Point Pilot Release.
- Containerized all Kotlin applications and developed Kubernetes manifests for Deployments and CronJobs.
- Developed the first centralized monitoring and logging platform for One Point using FluentD, Kafka, Logstash, and ELK.
- Built shared Helm charts and Groovy libraries to standardize CI/CD workflows across multiple projects.
Aspect Software, Inc.
May 2014 – May 2020
Engineer QA → Senior Software QA Engineer → Principal Automation Engineer
Dalian, China
- Grew from manual QA to Principal Automation Engineer over 6 years, building CI/CD and test infrastructure from the ground up.
- Designed a scalable Jenkins-based automation framework integrating 100+ QA suites with real-time reporting via ReportPortal.
- Built CI/CD pipelines integrating Jenkins with AWS, Bitbucket, Slack, Datadog, and Sumo Logic; provisioned test environments via Terraform and AWS Lambda.
- Developed automated API tests using C# and Selenium; built logging and monitoring pipelines using Datadog, Sumo Logic, and Grafana.
Interests
Education
Dalian University of Technology
2008 – 2013
B.A. Chinese Language and Literature
Self-taught in engineering and platform architecture.
Languages
RussianNative
Chinese (Mandarin)Native or Bilingual
EnglishFull professional