Marek Hlobil

A Python-based API for searching for flight, bus, and train tickets.

I was a member of the back-end team that was responsible for looking up the most competitive tickets and joining them together so the customer could get a door-to-door experience.

Part of the functionality was to compose unique itineraries by connecting flights (but also trains and buses) from different carriers and providers that were otherwise sold alone. That was the unique feature of Kiwi.com called virtual interlining. Another part of the process was filtering flights by rules stored in the database and pricing them according to dynamic rules.

The API communicated with multiple services to get content and additional metadata.

This site is a news site that is one of the most visited in the Czech Republic. It is also a part of Seznam.cz's portfolio.

I was responsible for the architecture of the API and redaction system. We used MongoDB as a database and the rest of the site was written in Flask with the front-end in React. Part of the redaction system was ab article WYSIWYG that was generating JSON instead of plain HTML. We also used Atomic design methodology. Thanks to this approach, the system was easy to modify, had backward compatibility, and was easy to maintain.

This is a site that deals with sports news and is a part of Seznam.cz's portfolio.

I was responsible for the redaction system and back-end for the website. The core part of the site was written in Django. The redaction system was written in a way that it was possible to use it on other sites (e.g., Super.cz).

This is a site about game reviews. The first service we deployed was with Docker.

My responsibility was to build a redaction system and back-end; both were written in Flask and shared a data layer was using MongoDB.

Lide.cz is dating site within Seznam.cz's portfolio.

My responsibility was to design the back-end architecture and implementation. We used Python and SQLAlchemy for data access. We also implemented support for real-time chats with WebSockets. We needed to scale WebSocket connections to multiple Nginx servers and used Redis Pub/Sub to distribute messages from the back-end to clients connected to Nginx.