Michelle Tsacheva

Created Greenfield platform for mixed load with AWS and on-prem. Co-led the project observability, service mesh, and continuous integration. Collaborated as a part of the senior engineers' committee to help guide the team's technical direction.

Created network architecture changes to create private subnets and restrict access to the minimum required operation. Redesigned security groups to restrict access to minimum required. Added minimum permissions AWS SSO (single sign-on) group permissions. Implemented AWS RDS database access via temporary IAM credentials. Added secure access to the company's infrastructure via managed VPN and SSH bastion host for the BI system.

Migrated Kubernetes from Heroku to Amazon EKS as part of my role. Deployed a Ruby on Rails service in a Heroku backward compatible way and auxiliary services to Kubernetes using CircleCI and TeamCity. Automated an AWS infrastructure with Terraform and implemented automation with Python and Shell.

Bootstrapped the observability team in Schibsted, taking over services from previous owners, defining operational procedures, and making initial service automations. Created the onboarding procedure for the observability team and onboarded and mentored eight DevOps and SRE engineers. Created automated provisioning for the metrics system to AWS accounts and Kubernetes namespaces. This kept us from making snowflakes in over 300 accounts and namespaces. Afterward, it was abstracted and reutilized for ten services by three teams. Designed and implemented a high-availability multi-AWS region metrics and alarming system. Built an AWS cost allocation and optimization platform. Improved the Schibsted AWS account bootstrapping process. Enabled the search engineering team and Subito in Italy, helping the teams onboard the standard Schibsted platform services, architecture, technical advice, and mentoring.

Improved load speed of slow pages up to 90% and 50% on average (Q4 shopping season availability).

Automated a provisioning tool for an SNMP monitoring system. It saved approximately eight hours of manual engineer work per install with about 200 installations scheduled in the following six months and used also for base lining manually provisioned installs. The system was designed to do everything from partitioning the disk, bootstrapping the system, and installing the deployment system client to post install validation and updating dashboards after the new install. It supports remote hands mode for high security networks—GovCloud for the US Department of Defense and AWS Secret region for the US Central Intelligence Agency. Streamlined tools for provisioning and managing the SNMP monitoring system. Built new, improved, and aggregated metrics and dashboards for SNMP monitoring system health. Provisioned tool support for deployment using remote hands in high-security networks—ITAR (US Department of Defense) and private customer cloud networks (Central Intelligence Agency). Built a system for managing discovery of network devices—daily rediscovery, priority (re)discovery, and device credentials rotation. Performance tested and optimized a new SNMP monitoring system. Communicated with third-party vendors of the SNMP monitoring system—technical and management.

Developed data-center operations metrics and tools. Owned an internal AWS dashboard for infrastructure teams. Upgraded Amazon retail DNS, excluding Route 53, software, improving automation, and executing key upgrades.

Enabled fully automated provisioning of new customer websites in order to allow the marketing of a free product edition that requires no manual work at all. Automated bandwidth accounting and quota cut offs or bandwidth surcharge depending on package. Worked as part of a team, implemented the API calls and back end but not the UI.