Tasks performed at ContentSavvy, Janya and Cymfony:Implemented a high-performance finite state transducer natural language processing pipeline and associated grammar compilers and executorsDesigned and implemented a distributed processing architecture that enabled the company to scale its natural language processing capability using CORBA and SOAPResponsible for performance tuning, benchmarking and comparison of MySQL and MongoDB databases enabling storage of up to a hundred of million documents per day in MySQLResponsible for several iterations of performance optimizations for 500,000 lines of natural language processing code resulting in increases of speed ranging from 50% to 200%Implemented Unicode in 500.000 lines of code, enabling the creation and use of resources for languages such as Chinese and ArabicMade natural language processing pipeline thread-safe for multi-core and multiprocessor computersImplemented document retrievers and converted resulting JSON and XML documents to a unified format for diverse sources, such as WordPress, YouTube, Facebook, LexisNexis, Factiva and DialogImplemented custom document storage for incremental updates and retrieval of millions of documents, query programs and a C++ API, facilitating linguistic research and developmentDeveloped a system for customizable database schema creation and corresponding text extraction definitionDeveloped a cross-platform wxWidgets desktop application for showcasing the natural language processing pipeline capabilitiesDeveloped a .NET desktop application for browsing extracted information stored in databasesDeveloped a text zoner tool for customizable markup of text files based on regular expression patterns, extensible with custom proceduresDeveloped a chat monitoring system for processing real-time data streams from IRC with alert channel notificationsPorted the natural language processing pipeline and related source code and build environment from Linux to Windows

Designed and developed a client/server system for communication between X Windows and text client programs and remote imaging systems consisting of both desktop control computers and custom hardware control systems.