Senior Platform Engineer | DevOps Engineer | SRE

Architected and governed multi-region, multi-account Kubernetes platform serving as Internal Developer Platform for 30+ teams at 99.99% availability. Built 4-account AWS landing zone (Organizations/Control Tower/SCPs) in Terraform, cutting provisioning from 5 days to 2 hours (95% reduction) and standardizing it enterprise-wide. Implemented multi-region EKS cluster/workload account separation, minimizing blast radius and simplifying IAM policy management across 30+ production, staging, and dev teams. Built GitOps platform with ArgoCD, Helm, and blue/green deployments, enabling self-service and increasing deployment frequency 3× (bi-weekly to daily) while cutting incidents 70%, mentored teams on adoption. Replaced static AWS credentials with IRSA for per-service-account least-privilege access, improving security posture and enabling full CloudTrail auditability for SOC 2 Type II. Enforced DevSecOps via policy-as-code (OPA + Checkov + tfsec) in CI pipelines, blocking 100% of non-compliant changes and reducing high/critical CVEs by 80% through Trivy scanning and KMS encryption. Engineered comprehensive SLI/SLO observability stack (Prometheus + Grafana + Loki + OpenTelemetry), decreasing MTTR 35% (85 to 55 minutes) with proactive anomaly detection. Drove FinOps program (Reserved Instances + Spot + rightsizing), delivering $1.2M annual savings (25–30% reduction) with zero impact to reliability or performance, including cost allocation and chargeback models. Established EKS upgrade governance (disruption budgets + staged rollouts + PDBs), achieving zero-downtime upgrades across all clusters in line with Well-Architected reliability practices. Standardized secure pod-to-AWS access patterns using IRSA and RAM subnet sharing, enabling least-privilege workload isolation in multi-account environments.

Owned infrastructure and reliability for Kafka-based Centrifuge event delivery platform processing 2B+ daily events at 500K+ peak throughput with 99.9%+ availability. Architected AWS foundation (EC2, VPC, MSK, RDS) for Centrifuge distributed job scheduler, reliably delivering billions of events to hundreds of partner APIs while absorbing transient failures. Migrated 200+ EC2 instances to modular Terraform modules, reducing provisioning time from days to 30 minutes and establishing IaC standards adopted across the organization. Designed Kafka lag + CPU-driven autoscaling for Directors (ECS), handling 4× traffic spikes and reducing cloud spend 20% through efficient resource scaling. Engineered per-tenant retry isolation and backpressure, preventing cascading failures during partner outages and preserving global delivery SLAs for high-cardinality workloads. Built end-to-end observability stack (Prometheus + Grafana + ELK + Datadog), slashing mean time to detection 80% (25 to 5 minutes) and accelerating incident triage. Led post-incident analysis and remediation processes, implementing job state machine improvements (exponential backoff, archival) that drove systemic reliability gains.

Contributed to backend services and monolith-to-SOA migration supporting global-scale booking traffic with safe, incremental rollout practices. Executed monolith-to-SOA migration using dual-read gating (1% → 100% ramp), Diffy comparisons, and canary promotions, eliminating rollout incidents during decomposition. Built services with Thrift IDL and generated RPC clients (mTLS, retries, context propagation), reducing cross-language drift and accelerating multi-language development. Operationalized services with standardized metrics, templated dashboards, and IDL-defined alerts (p95/p99 latency, error rate, QPS), significantly reducing time-to-production readiness. Developed high-performance REST APIs (Java/Spring Boot/PostgreSQL) achieving sub-100ms P99 latency under 50K+ concurrent requests. Reduced peak API latency 25% by introducing asynchronous Kafka workflows and Redis caching, enhancing system resilience and user experience.