yorko / deperson Goto Github PK

• 🔭 I’m currently working at RELX as a Principal Machine Learning Scientist in Research Integrity
• 🤓 I’m leading mlcourse.ai – an open Machine Learning course
• got a Ph.D. degree in applied mathematics
• I'm blogging about machine learning, mathematics, quantum computation, career development, programming, soft skills, popular science and anything else that I find exciting
• can also brag about my Kaggle Competitions Master title, Kaggle profile
• my team won the 1st place in Google's Kaggle competition on natural text comments classification
• YouTube channel views:

Born in Russia, made a world around living 4 years in Israel and 1.5 years in Canada. Then back to Russia. Currently residing in the Netherlands.

• 2007 - 2013. Studied aviation and later Computer Science at Moscow Institute of Physics and Technology
• 2011 - 2013. During Master's, worked as a Business Intelligence architect, switched to full-time Ph.D. studies cause life is too short
• 2018. Got a Ph.D. degree in applied mathematics at Higher School of Economics, Moscow
• Jan 2017 - Jun 2018. Worked as a Data Scientist @ Russian IT giant Mail.Ru Group
• Jun 2018 - Apr 2020. Worked as a Data Scientist focusing on NLP @ KPN, the Netherlands
• 2017 - curr. Leading open Machine Learning course mlcourse.ai
• Apr 2020 - curr. Working as a Principal ML Scientist focusing on NLP @ RELX, the Netherlands

mlcourse.ai is an open Machine Learning course by OpenDataScience (ods.ai), led by me. In 2017-2019, I've been leading active session of the course, offering essential theoretical ground and a ton of practice (assignments, Kaggle Inclass competitions, projects, etc.). All for free. Some 26k people participated, ~1500 finished the course, I got dozens of direct messages on how passing the course changed careers of mlcourse.ai alumni. Right now, the course in a self-paced mode, still has much to offer.

In a team project, I lead the development of a cryptonews sentiment barometer that scrapes daily news, scores them with a pre-trained BERT model and visualizes in a comprehensive way. It's built for fun, I don't believe you can predict BTC prises with these sentiments. We also don't plan to sell such predictions through an API (although this is possible) because them maintaining it all is not fun anymore. But I strongly recommend everyone to work in a team on a pet-project at some point, that's a great experience. I described it in Russian here.