Work Experiences

MD Fashion is a group of partner companies that have united more than 140 branded stores in Ukraine and Kazakhstan: Tommy Hilfiger, Diesel, Gant, G-Star Raw, Replay, Walker, STEM, Under Armour, MEXX.

After the company decided to independently handle the development and operation of the website, moving away from outsourcing services, I was the first to join the team.

Later, I participated in forming and training a team that was also transitioned from the internal staff. During this time, we implemented a wide range of changes. We integrated APM monitoring, OWASP, and achieved fault tolerance. Subsequently, I implemented the full Infrastructure as Code (IaC) deployment cycle, and later fully transitioned to Kubernetes. Currently, we manage six clusters.

To improve our development velocity and reduce toil, I build automated pipelines for our services to run tests and linting for each commit merge request and to automatically release and deploy the services once they are merged. During the first year we got started with Bitbucket Pipeline and bash script but then switched to ArgoCD because bash script needed too much maintenance and we liked the Bitbucket Pipeline/ArgoCD colaborate better.

MD-FASHION is a group of partner companies that have united more than 140 branded stores in Ukraine and Kazakhstan: Tommy Hilfiger, Diesel, Gant, G-Star Raw, Replay, Walker, STEM, Under Armour, MEXX.

When I joined MD-FASHION, I initially focused on the ESXi, which was support to the main business software, MSSQL, networking, etc.

Later on, my role at MD-FASHION shifted and I started working more on infrastructure and DevOps topics. I planned and implemented a new development platform for our growing team on top of Docker. To sustainably learn from past issues and avoid being waken up by alerts, I integrated After-action review, which helped us fully explore incidents and solve the root causes.

To improve our confidence and velocity during development and deployment, I implemented the full CI/CD lifecycle for all of our services, using Jenkins and finally an automated release to Production.

To help us debug any incidents and investigate for customer support tickets, we used internal logic 1C system, but this proved to be too expensive in the long run. I developed our alternative monitoring and observability stack using ELK, Grafana, and Prometheus. Using these, I also established SLOs for one of our core metrics, “time to export events”.