Lecturer on Big Data and industrial development on python courses.

Created an infrastructure for production application of machine learning models.Transferred retail products optimization process on new infrastructure.

Developed a microservice with machine learning for the Moduze mobile app.The microservice is written in Flask. The main framework for machine learning tasks is Pytorch.During my work as a Data Scientist:- I followed sota solutions and implemented the ideas described in them;- Made analytics for improving data for training;- Communicated with customers, planned team sprints;- Thought over the architecture of integration with backend;- Wrote a microservice in Flask;- Created autocomplete, next word prediction model, model for image retrieval;- Made integration with the main backend of the application;Microservice solved the following tasks:- Autocomplete, which was offered to the users based on their history of his requests and available items in the database;- Parsed the search query to simplify the search in PostgreSQL;- Searched for a similar product by photo (Pytorch + redis + hnswlib);- Was engaged in the deployment of infrastructure on the local server of the company- Moved machine learning microservice to docker-compose for production. Helped the devops engineer to port this to kubernetes.

During my work as a backend developer:-Developed a backend for a mobile application on Django rest. I wrote an api for the client;-Worked on transferring logging to the ELK stack;-Produced integration with our ether blockchain node using web3py;- Moved application deployment to docker-compose;- Worked according to the SCRUM methodology

Тестировщик ПОDuring my work as tester at part-time:- Was engaged in the development of automatic testing of the api platform;- Wrote scripts of users behavior;- Conducted regression testing;

Worked in two laboratories.In the laboratory of neuromorphic systems, i studied the basics of machine learning. I was engaged in reinforcement learning tasks based on the example of Atari games. Implemented the correlated dropout algorithm on tensorflow on the MNIST handwritten digits problem.At the Keldysh Institute of Applied Mathematics of the Russian Academy of Sciences, i was developing a microservice in the GO language, which was responsible for the parallel reading of data obtained by the molecular dynamics method with subsequent processing in Python. The processing consisted in drawing the areas needed by the researchers for analysis.