Inkvizitor Platform
A production platform used by thousands: Kubernetes infrastructure, backend, data, observability, and delivery treated as one system.
16k+users
3 AZinfrastructure
3–4kactive users
~10 mindeploy
01 / OVERVIEW
Context
Inkvizitor is a large Telegram / web platform with many internal services and a real active audience. My work spans infrastructure and backend: from networking and Kubernetes to data, secrets, monitoring, and release delivery.
The main engineering goal was to make a complex production system calmer: remove manual steps, create a repeatable path for change, and make the platform observable.
02 / SCOPE
My role
- Production Kubernetes and multi-AZ infrastructure
- CI/CD and GitOps delivery with GitLab CI/CD and Argo CD
- Python backend and internal services
- PostgreSQL, Redis, ClickHouse, and RabbitMQ
- Vault, ingress, object storage, and observability
03 / OUTCOME
Outcome
- The path from change to production dropped from roughly two days to about ten minutes.
- A manual delivery process became a repeatable CI/CD and GitOps flow.
- Infrastructure, backend, and observability became one connected product system.
STACK
Technologies
KubernetesPythonPostgreSQLRedisClickHouseRabbitMQVaultArgo CDGitLab CI/CDGrafanaLoki