Skip to content

MarioPrskalo

Software EngineerJava, Kafka & React

I build high-traffic, event-driven backend services and the React interfaces on top of them.

Europe (CET)

GitHub (opens in a new tab) LinkedIn (opens in a new tab)Email

About

I'm a software engineer at Sportradar. I studied computer science at the University of Mostar, where my thesis was FAInance, a React financial dashboard with its own forecasting model. I came into the industry through NSoft's infrastructure bootcamp.

Outside work, I'm building invoice-relay and writing about what I learn, starting with backend-for-frontend security and the transactional outbox.

Education
BSc Computer Science, University of Mostar2020–2024
Languages
Croatian (native) · English (C2) · German (A2)

Experience

  1. – Present

    Junior Software EngineerSportradar

    One of five engineers on the team that owns high-traffic, revenue-critical services.

    • Build and run event-driven services in Kafka and Quarkus, and the React screens that sit on them.
    • Migrate legacy services to Kubernetes and AWS with Terraform, with metrics, tracing and alerting in Prometheus, Grafana and OpenTelemetry.
    • Led the redesign of an internal tool’s UI and UX, and reworked its backend workflows to increase capacity.
    • Java
    • Quarkus
    • Kafka
    • React
    • TypeScript
    • AWS
    • Kubernetes
    • Terraform
    • OpenTelemetry

  2. –

    Infrastructure Bootcamp (Trainee)NSoft (acquired by Sportradar)

    One of 20 trainees selected from over 100 applicants; offered a full-time role before the halfway point.

    • Linux
    • Docker
    • Kubernetes

Projects

  • In progress

    invoice-relay

    A small invoice pipeline built around one question: how do you make sure an invoice is never sent twice? The answer will be a transactional outbox, idempotent consumers and a test that delivers the same event twice.

    • Spring Boot
    • Kafka outbox
    • React
    • AWS
    Fig. 1 — Planned flow
    1. APIhttp
    2. Outboxdb table
    3. Kafkatopic
    4. Consumeridempotent

  • BSc thesis

    FAInance

    Source (opens in a new tab)

    A React financial dashboard that forecasts revenue with an LSTM model and linear regression.

    • React
    • TypeScript
    • Node.js
    • Express
    • MongoDB
    • LSTM

Writing

  • Upcoming

    Securing a backend-for-frontend

    Keeping tokens out of the browser: cookies, CSRF and per-tenant checks.

  • Upcoming

    Idempotent consumers and the outbox

    Why “exactly once” still needs your help.