Gatra – Kubernetes, Terraform, Golang

A personal project focused on open-source tooling for CI/CD and container-based development, built using Rust and Go. The platform provides ready-to-use GitLab components and deployment utilities, simplifying the building, testing, and shipping of Go projects while supporting modern DevOps workflows.

• Founded and actively maintain Domaine Gruppe, an open-source collection of reusable GitLab CI components designed to standardize and accelerate CI/CD pipelines;
• Designed and implemented ready-to-use CI/CD workflows for Go, Rust, and React Native applications, including build, test, containerization, and deployment stages;
• Integrated Semgrep SAST/SCA into pipelines to enforce secure-by-default development practices;
• Built automation for container image building and deployment, enabling consistent, reproducible releases across teams;
• Focused on developer experience by reducing CI/CD boilerplate, improving documentation, and enabling rapid onboarding for new projects;
• Provided Docker images and standalone tooling to support teams not using GitLab directly;
• Promoted CI/CD best practices and GitOps-friendly workflows, helping teams improve delivery speed, reliability, and security.

Migration of a production infrastructure from Docker Swarm to Kubernetes (K3s) to enhance scalability, reliability, and operational consistency. The platform supported multiple business-critical backend services, requiring high availability, secure deployments, and clear audit trails. The migration introduced a GitOps-first approach with FluxCD, making all infrastructure changes fully declarative, versioned, and traceable.

• Owned end-to-end migration strategy and execution from Docker Swarm to Kubernetes, including cluster design, service refactoring, deployment automation, and rollout planning;
• Replaced manual and Swarm-based workflows with declarative Kubernetes manifests, reducing configuration drift and deployment errors;
• Introduced GitOps workflow with FluxCD for fully versioned and traceable application and infrastructure deployments;
• Implemented SOPS-encrypted secrets with secure key management for safe Git storage;
• Standardized Helm-based deployments and enforced environment separation;
• Integrated observability and logging to ensure production readiness;
• Improved security posture, auditability, system stability, and enabled faster, safer releases.

A new loan calculation engine to replace a legacy system that relied on approximation-based logic and produced inconsistent financial results. The engine serves as a core backend service for calculating installments, interest, and repayment schedules across multiple loan products. The project focused on improving calculation accuracy, enhancing product flexibility, and ensuring long-term maintainability to support business growth.

• Led full technical design and implementation of a new loan calculation engine;
• Refactored core calculation logic to ensure financial correctness and consistency across products;
• Identified and fixed critical inaccuracies in the legacy engine causing mismatched results;
• Implemented a configuration-driven model in PostgreSQL, enabling rapid launch of new loan products without redeploying services;
• Designed a clean, modular backend with clear separation of domain logic;
• Ensured backward-compatible integration with existing services;
• Improved trust in financial data, reduced regression risk, and supported faster business experimentation and revenue growth.

A modular notification backend service in Golang to deliver critical customer communications, including password resets, loan reminders, status updates, and marketing messages. The service supported a fintech platform for active loan customers and internal product teams. Emails were sent directly via SMTP to reduce costs and maintain full delivery control. Templates were dynamic, built with Go templates, and stored in PostgreSQL to allow rapid updates without code changes.

• Designed and implemented the notification service architecture, including API design and data modeling;
• Refactored the system using clean architecture and dependency injection to simplify adding new channels;
• Built a reusable template system for multiple message types and channels;
• Delivered emails directly via SMTP, maintaining full control over delivery and reducing costs;
• Ensured scalability and reliability of the service to support growing notification needs;
• Improved development velocity and operational efficiency by standardizing core logic and templates.