The services were delivered to the UK-based company with a focus on delivering relentless protection for their clients. The core product under design and development was a tool that was capable of visual identification if a web page screenshot was a login page. The tool comprised a very problem-specific, custom deep-layer structure (40 layers), that contained only a dozen of trainable parameters. The current version of the tool had 99% sensitivity, 80% specificity, and 800ms of processing time.

The project under development was an open-source domain-specific language written as the extension of CMake that allowed a new, user-friendly paradigm of writing CMake code. It targeted medium to large CMake deployments and helped by facilitating modularization and parametrization of the project description. The extension aimed to be a more user-friendly way of writing complex CMake build systems than the bare CMake, freeing a mind for actual programming.

The client was a Polish Institute of Meteorology and Water Management. The main goal behind the project presupposed setting the framework for porting the existing Fortran code base that implemented the dynamic core into the C++ using the specialized libraries written for GPU. Besides coding in C++, I introduced code versioning, project management and unit tests to the team.

The main goal of the cooperation was design of the Ontology-based Web Database ffor understanding the Amyotrophic Lateral Sclerosis.My responsibilities includedbuilding the offline version of the database, design and execution of all statistical inference.

The client, one of the leading Polish banks, needed assistance with the development of a scoring model for selecting customers for cross-sell campaigns using SAS and Oracle. Developed features included SQL queries dependency tracking, simultaneous execution of queries, capability to perform statistical procedures and creation of automated reports, modular and easily extensible architecture.