Csaba Keri

With another developer, we made a recursive engine that could build and serve web pages based on a recursive XML declaration with its own nodes, page processing, and building engine, which filled the sites with multilingual content (taken from the disk, session, and SQL server).

The engine could implement a graceful degradation and keep the core functionality even if all the auxiliary components failed. The logic contained a continuous performance monitoring system to detect the bottlenecks and DDOS attacks immediately.

I was the head of back-end development on an on-demand TV channel with monthly subscription fee. My responsibility was to ensure bug-free and architecturally optimal code and manage the releases twice a week in more than 15 countries on more than 50 serves. Besides being a lead, I took part in the code development on the back-end side and the client side as well—I wrote a client for a set-top box in Javascript. I took part in the weekly planning meetings and the management to have the best roadmap and deliver the highest possible business value from week to week. I was also responsible for the back-end API, which was used by all of the clients (Win Mobile, iOS, Android, and web browsers).

A C# and SQL server-based, heavily parallel, distributed application with very complicated calculations on a large amount of data (100,000+ data points/customer) with strong performance expectations (regarding the number of the calculations/hour).

My contributions were the data loading (extract, transform, load), the parallel logic in the business layer, several rules, and performance optimization with remarkable results. I was also responsible for the quality assurance of the imported data and the detection of inconsistencies, missing data points, and everything else.

I took part in the core architecture planning and was the only person responsible for the core application part, forming the basis for the customized app features.

Also, I maintained the (virtual) hardware environment (including the Windows, SQL servers, and Docker). I kept the application stable and alive at any circumstances (because of the strict legal delivery deadlines).

Further, I created an end-to-end testing tool, which was able to compare any parts of the reports and find all the differences, regressions, unforeseen data constellations, and calculation errors with a speed of 3,000+ reports/hour.