Radoslav Terezka

Jobiqo is a job board platform initially built on Drupal 7, which required a complete redevelopment on a newer Drupal version in which I played a fundamental role.

I was a vital part of the back-end development for this completely new version based on Drupal 8/9, React, and GraphQL. There I ported plenty of features from the former version to a new one, came up with many architectural ideas and improvements, and wrote many GraphQL queries and mutations.

I configured the virtualized local development environment using the DDEV tool and set up the testing infrastructure using the Behat testing framework. My work also included configuring CircleCI pipelines and Jenkins jobs, writing lots of Bash and Python automation scripts, setting up Solr configurations, preparing several production environments, and assisting with launches by guiding my team members through many tasks.

Contentpool was a Drupal 8 distribution that enabled replication of content from one main instance into satellite sites by using the CouchDB Replication Protocol under the hood.

I helped to shape the distribution and a client project using it with the following achievements:

• Established a common data model to be shared for both main and satellite sites.
• Solved several problems regarding content replication to satellite sites when content is added, edited, or deleted on the main site.
• Developed a rendering strategy for the satellite sites to render entities distributed from the main site.
• Made several public modules during the development to cover distribution use-cases (the markup field, custom elements, Basic Auth Limited, trusted redirect, entity edit redirect).
• Wrote a couple of behavioral tests using the Behat framework to cover crucial replication behavior.

Multi-site Food Digital was a Drupal 8 project built on a multi-site architecture for a large German publishing company covering five different sites.

During development, I established the configuration strategy that the correct configurations and settings are used per environment and also per site.

I ensured that it was the correct Solr configuration with all required fields and processors used in different environments. On production, I set up Acquia Connector to be correctly integrated with Acquia Search service.

Regarding CI/CD, I worked a lot with Travis CI where I learned how to use it, debug and I also improved our Travis CI configuration. The production was run in Acquia Cloud, where I set up pipelines with all the necessary hooks to correctly deploy the application to production.

I was also involved in introducing more efficient caching strategy by establishing custom cache tags and invalidation logic for them. I also gained an experience in CloudFlare and its purging options. Regarding quality assurance, I wrote a lot of behavioral tests using the Behat framework.

A special ELK stack for the client allowing not only to parse out different log types from different servers but also to consume Drupal logs from a client's Drupal application.

My part on the project was to develop a Docker Compose file bringing up whole ELK with additional supportive containers:
• Configured NGINX proxy container to make Kibana and Logstash' consuming HTTP endpoint available to outside world under desired paths on a specific domain protected by basic authentication.
• Set up a subgroup of containers allowing to generate Let's Encrypt SSL certificates for the domain where the stack was run.
• Developed a complex Logstash config file with proper parsing patterns and mutations allowing to consume different types of logs from files and also consuming Drupal logs via HTTP.
• Established multiple dashboards, searches, and visualizations for Kibana to be imported out-of-the-box whenever and wherever the stack is run so they don't need to be created manually.
• Introduced a special container allowing SSH so the different servers can regularly push the different log files for the Logstash to be consumed.
• Adapted Drupal's logs_http module to align with Logstash and make the consuming of Drupal logs via HTTP work.

Kurier is a large daily newspaper based in Vienna.

The company that I was working at the time was responsible for delivering a fresh new editorial back-end based on Drupal 8 and Thunder distribution. It was also acting as a content pool (used headlessly) to provide REST endpoints for various front-ends based on AngularJS 2. Additionally, Drupal 8 editorial back-end was connected to Retresco to index and query the content.

Besides being a developer, my role in the project was also as a team lead and solution architect (acting as an interface between the client and my team, discussing requirements, and finding solutions). I also prepared a fully virtualized local development environment built by Vagrant and provisioned by Chef, which was used by our own team and the client. I was also partially responsible for setting up the CI/CD with Jenkins.

From the development point of view, I implemented plenty of REST endpoints, made multiple migrations from the old system running on-the-fly, integrated search API views, and queried Retresco. I also wrote plenty of serializers to serialize Drupal entities to Retresco documents, and last but not least, I attempted to swap the default SQL storage to Elasticsearch.

An interesting little internal project for a large publishing company that was monitoring the performance of social networks.

I refactored the project into a Symfony command bundle because it malfunctioned. Before starting, I wrote a couple of Chef cookbooks to provision the whole setup. The refactoring output was a Symfony command bundle regularly triggered by a Cron job to perform regular information fetching.

An ambitious international project built from scratch with a Headless Drupal 7 back end, AngularJS front-end, Prerender.io, Solr, and Varnish.

First of all, I was responsible for setting up the Solr environment with all field types and their corresponding analyzers. Then I implemented plenty of migrations running daily between various systems, sometimes importing up to 2,000 recipes per day. I also wrote a couple of feed imports to import recipes from various feeds (Atom, RSS). The most interesting part of the project from my side was to set up a Drush daemon to run constantly to process multiple tasks from the queue, like invalidating the Varnish cache, triggering a Prerender fetch, indexing to Solr, generating multiple image derivations for our CDN, and many more. During the project, I also adjusted our deployment shell scripts.

This was a relaunch of an online magazine from an old system to Drupal 7, which was in French.

Here I gained my first Solr experience, where I established the Solr environment and implementing an autocomplete, full-text, and faceted search. The search was in French, which required me to set up stemming as well. I also migrated a large amount of content (a complicated structure and users) from the old system to Drupal 7. The most interesting task during development was setting up Varnish caching for authenticated users (using the Authcache module and a specific Varnish setup). Finally, I was involved in performance tuning by using the XHprof profiler to identify bottlenecks and improve the overall performance.

A gamified financial platform based on Drupal 7 for a Swiss company called eny Finance which provides lending solutions in Switzerland.

Here the site users could actively contribute to the content and gain virtual currency called Talers for multiple actions. The currency could also be spent.

I implemented several features on the site:
• Created an idea competition where registered users could post ideas about new financial products, those ideas could be voted, and the idea owner with the most votes won a price.
• Made several adjustments to the registration/profile, where users could gain Talers for filling in various profile information (profile picture, setting up a strong password, and more).
• Implemented an online chat for a financial expert to enter and interact with various users who ask him questions. The chat history was completely stored to be accessible for those who could attend.
• Extended Adform tracking for various user actions like entering the chat, voting for an idea, and others.