Platform modernizations
What I replaced, and what runs in its place now.
FINSIGHT
T-REX platform (acquired by FINSIGHT, 2024) · Jul 2022 – Present · all of it initiated and done by me- InfrastructureAnsible-managed VMsreplaced byTerraform on AWS
- RuntimeMonolithic services stackreplaced byAbout 20 microservices per environment, scaled by HPA
- BuildsAMI builds, up to 25 minutesreplaced byContainer builds, about 10 minutes
- DeliveryAzure DevOps pipelinesreplaced byGitHub Actions and ArgoCD GitOps, blue-green with Argo Rollouts
- TestsRegression tests on ECSreplaced byOn Kubernetes with Karpenter, KEDA and HPA, about 70% cheaper per run
- ObservabilityNew Relicreplaced byOpen-source LGTM stack: Loki, Grafana, Tempo, Mimir, about 75% cheaper
- AcquisitionSeparate T-REX and FINSIGHT infrastructurereplaced byAccounts, networking and SSO integrated
OCTOPOL
sole DevOps engineer · VMware vSAN, AWS and GCP · Mar 2021 – Jul 2022- ServersStandalone servers, including about 8 TB of databasesreplaced byVMware vSAN cluster: 10 physical hosts, up to 100 VMs, minimal downtime
- ResilienceFrozen servers restarted by handreplaced byVMs fail over and migrate between hosts automatically
- CapacityEach server limited to its own hardwarereplaced byPhysical cluster expanded on demand
- ContainersApps on standalone serversreplaced byKubespray Kubernetes (Calico, MetalLB) on vSAN, EKS on AWS
- MonitoringNo cluster monitoringreplaced byPrometheus
Experience
Jul 2022 – Present
DevOps Engineer
FINSIGHT (May 2024 – Present) · T-REX (Jul 2022 – May 2024), acquired by FINSIGHT
- Sole DevOps engineer for the T-REX platform, and on call for it: about 20 microservices on EKS across several environments, used by 50–60 developers, plus the Airflow and BI platforms.
- Migrated production from VMs to EKS while it stayed live. Split the monolith into microservices, wrote the Helm charts (later merged into one shared chart) and set up HPA per service. Builds went from up to 25 minutes to about 10.
- Switched all infrastructure from Ansible to Terraform and wrote our own EKS module with Karpenter.
- Handle all upgrades, including EKS and Kubernetes versions and cluster add-ons.
- Introduced GitOps with ArgoCD and moved all CI/CD from Azure DevOps to GitHub Actions. We release twice a week with blue-green deploys through Argo Rollouts, and hotfixes go out whenever needed.
- Took the infrastructure through SOC 2 audits at both T-REX and FINSIGHT: private EKS clusters behind zero-trust access, centralized secrets, keyless OIDC from CI to AWS and Dependabot. Implemented the controls and prepared the evidence.
- Merged the T-REX and FINSIGHT infrastructure after the acquisition: AWS accounts, networking, and SSO through IAM Identity Center with per-team permission sets.
- Replaced New Relic with self-hosted Grafana, Loki, Tempo and Mimir (LGTM), cutting observability costs by about 75%. Alerts go to PagerDuty.
- Run PostgreSQL on RDS in all three environments: upgrades, replication, migrations, cross-region backups and a yearly disaster-recovery test.
- Ported the regression tests, which run on every pull request, from ECS to Kubernetes with Karpenter, KEDA and HPA, cutting the cost of each run by about 70%.
- Help developers and data scientists track down performance problems.
Stack: AWS (EKS, RDS, ECS, IAM Identity Center), Kubernetes, Helm, ArgoCD, Argo Rollouts, Karpenter, KEDA, Terraform, GitHub Actions, Grafana, Loki, Tempo, Mimir, PagerDuty, PostgreSQL, Airflow
Mar 2021 – Jul 2022
DevOps Engineer
OCTOPOL
- Sole DevOps engineer across AWS, GCP and an on-prem VMware vSAN cluster, and on call for all of it.
- Moved standalone servers, including about 8 TB of databases (largest about 1 TB), onto a 10-host vSAN cluster running up to 100 VMs, with minimal downtime.
- Frozen servers used to be restarted by hand. Now VMs fail over and move between hosts automatically, and the cluster grows by adding hosts.
- Set up EKS for production and development on AWS, and self-managed Kubernetes on vSAN with Kubespray, Calico and MetalLB.
- Introduced Prometheus for cluster monitoring.
Stack: VMware vSphere and vSAN, AWS (EKS), GCP, Kubernetes, Kubespray, Calico, MetalLB, Prometheus
Feb 2016 – Mar 2021
Linux System Administrator
SPD Hosting (Mar 2019 – Mar 2021) · Host&Found (Feb 2016 – Mar 2019)
- Ran production Linux at an Israeli hosting and managed cloud provider: thousands of servers and websites, including large e-commerce stores where downtime meant lost sales.
- Worked with Magento, custom e-commerce stacks and the major hosting control panels, and managed clients’ AWS accounts.
- On call for production incidents and the first point of escalation for clients, handling each case through to the fix.
Stack: Linux, Bash, Magento, hosting control panels, AWS
Technical skills
- Cloud
- AWSEKSEC2ECSVPCS3RDSLambdaSQSSNSCloudFrontAPI GatewayElastiCacheOpenSearchEFSFSxGCP
- Kubernetes & GitOps
- Kubernetes (EKS and bare metal)HelmArgoCDArgo RolloutsKarpenterKEDACluster AutoscalerAWS Load Balancer ControllerNGINX Ingresscert-managerexternal-dnsKubesprayRancherCalicoMetalLBDocker
- Infrastructure as code
- Terraform (custom modules)Ansible
- CI/CD
- GitHub ActionsAzure DevOpsBlue-green deployments
- Observability
- GrafanaPrometheusLokiTempoMimirPagerDutyNew Relic
- Security
- SOC 2IAM Identity Center (SSO)OIDC workload identityAWS Secrets ManagerExternal Secrets OperatorZero-trust network accessDependabotHashiCorp Vault
- Databases
- PostgreSQL (RDS)MySQLMongoDB
- Virtualization
- VMware vSpherevCenterESXivSAN
- Data platforms
- Apache AirflowBI platforms
- Languages & OS
- Linux (10 years)BashPython
- Networking
- CloudflareDNS and CDNNetworking fundamentals (CCNA training)
Training and languages
- RHCSA/RHCE preparation course, John Bryce College (2017)
- MSSQL 2014 DBA course, John Bryce College (2016–2017)
- CCNA course, John Bryce College (2015)
- English proficient
- Hebrew fluent
- Ukrainian fluent
- Russian native