Andrey Vasnetsov
About
Detail
Co-Founder, CTO
Berlin, Germany
Résumé
Jobs
verified_user
0% verified
Co-Founder, CTOQdrantJan 2022 - Current (4 years 5 months)Qdrant is a vector similarity engine. It deploys as an API service providing search for the nearest high-dimensional vectors. With Qdrant, embeddings or neural network encoders can be turned into full-fledged applications for matching, searching, recommending, and much more!- Lead Data ScientistMoBerriesJan 2020 - Dec 2021 (2 years)Lead development of a new neural search based vacancy-candidate matching engine
- Software DeveloperMailRu GroupJul 2014 - Mar 2016 (1 year 9 months)C++ \Python programming Developed a task distribution system. Optimized algorithms for processing large data.
- Machine Learning Engineer \ Team leadTinkoffJul 2017 - Jan 2020 (2 years 7 months)Machine learning Tech Lead at Tinkoff.ru, Search department Headed a group of two Data Scientists and a Python developer. Led the development of a search project for goods and services. Built a machine learning platform, which allowed analysts to rapidly deploy their models in production.
- Software Engineerdotin An OpenSesame CompanyMar 2016 - Feb 2017 (1 year)Developed a social profile matching platform that is used to extract and process information about users of social networks. Analyzed open linked data sources, such as dbpedia.org, to enrich user data.
Education
verified_user
0% verified
- Master's degree, Information TechnologyBauman Moscow State Technical UniversityJan 2015 - Dec 2017 (3 years)Thesis work is on application of metric algorithms for sequence labeling and classification.
- Master's degree, Information TechnologyBauman Moscow State Technical UniversityJan 2015 - Dec 2017 (3 years)Thesis work is on application of metric algorithms for sequence labeling and classification.
- Bachelor's degree, Information technologiesBauman Moscow State Technical UniversityJan 2011 - Dec 2015 (5 years)Thesis: “Classification algorithm research for named entity disambiguation”
Projects (professional or personal)
verified_user
0% verified
- QQuaterionJul 2022Quaterion is a framework for fine-tuning similarity learning models. The framework closes the "last mile" problem in training models for semantic search, recommendations, anomaly detection, extreme classification, matching engines, e.t.c. It is designed to combine the performance of pre-trained models with specialization for the custom task while avoiding slow and costly training.
- QQdrantApr 2020Qdrant is a vector similarity engine. It deploys as an API service providing search for the nearest high-dimensional vectors. With Qdrant, embeddings or neural network encoders can be turned into full-fledged applications for matching, searching, recommending, and much more!
- MMention classifierMention classification lays between Named Entity Recognition and Entity Linking. It provides much more detailed information about an object than NER does. At the same time, it does not require storing and maintaining any knowledge base with known objects. The Classifier able to work with absolutely new objects never appeared in the train set.
Publications
verified_user
0% verified
- MSimilarity Learning lacks a framework. So we built oneMeduimcomJun 2022Similarity learning can overcome traditional problems of classical machine learning models, but it is mostly a domain of academic studies if we won't have convenient instruments.
- Neural Search TutorialMediumJun 2021Step-by-step guide on how to build a neural search service. - What is neural search? - Which model could be used? - What tasks is neural search good for? - Prepare data for neural search - Vector Search Engine - Make a search API - Deploy as a service
- Metric Learning Tips & TricksTowards Data ScienceMay 2021How to train object matching model with no labeled data and use it in production. - What is metric learning and why using it? - Data for Metric Learning - Training the Model - Model Confidence - Neural Rules - Vector Search in Production
- PFilterable approximate nearest neighbors searchPersonal BlogNov 2019How to make ANN search with custom filtering? Search for closest vectors only in selected subsets.
- aGeneralization of metric classification algorithms for sequences classification and labellingarXivorgOct 2016The article deals with the issue of modification of metric classification algorithms. In particular, it studies the algorithm k-Nearest Neighbours for its application to sequential data. A method of generalization of metric classification algorithms is proposed. As a part of it, there has been developed an algorithm for solving the problem of classification and labelling of sequential data. The advantages of the developed algorithm of classification in comparison with the existing one are also discussed in the article. There is a comparison of the effectiveness of the proposed algorithm with the algorithm of CRF in the task of chunking in the open data set CoNLL2000.