CEO at Logrus IT
Leonid graduated with honors from the Moscow Institute of Physics and Technology (MIPT), specializing in plasma physics and computer simulation. He started working as an engineer at the Moscow Radiotechnical Institute (MRTI).
In 1991, he joined one of the first ever software localization projects, which turned out to be a life-changer. In 1993, he co-founded Logrus and has served as the company’s CEO since then. He has been CEO of Logrus IT since 2016.
True to his scientific background, Leonid is continuing to research and publish papers on issues such as Language Quality Assurance methodology, process and metrics, client and project management, industry revelations and myths. His goal is to bring a hard-core, fundamental, objective and unbiased scientific approach and knowledge to areas traditionally occupied by project managers and linguists. Leonid is relentlessly challenging common knowledge, traditional ways and widespread prejudice, and enjoys working with people who demonstrate creative thinking.
MT Deployment: Best Practices
The presentation provides a brief overview of the current state of MT use (including major pros and cons), outlines notable market trends, and goes over various takeaways from the ongoing MT deployment epic at Logrus IT. We start with overcoming prejudices and then discuss MT support by various CAT tools and selecting the best engine(s).
We proceed with various approaches to MT quality evaluation, process specifics, and ways to address the topic of MT deployment with both clients and suppliers, including the crucial issue of MT-related discounts. We conclude with practical recommendations, including both pleasant surprises and unexpected pitfalls.
Cabbages and Quality: Creating Quality Metrics Applicable in Real Life
The presentation discusses various shortcomings of the currently published MQM and DQF quality approach and error typology and suggests a completely ready, fully scalable, 3D, hybrid Quality Triangle approach to measure translation quality. It then goes on to explain how existing error typologies can be modified to become truly scalable, how they are integrated into the Quality Triangle model, and how to create quality metrics of varying complexity on the fly using ready, "frozen" components.