Worked on a medical chat-bot project.
* improved data collection approaches for labeling using active learning, Snorkel, and pseudo-labeling;
* developed an approach to check the quality of text labeling, which improved the speed of analysis and helped to find errors in labels;
* trained ML models for text classification and NER;
* experimented with various approaches to improving model quality, including text augmentations, modifications to model architectures, different loss functions;
* developed a training pipeline based on PyTorch-Lightning and Hydra;
* introduced principles of writing good code including texts, style checking, autoformatting;
* mentored colleagues;
* organized weekly meetups to share expertise;

Worked on video enhancement project:
* using video super-resolution approaches to improve video quality;

- Implemented churn prediction, fraud detection and look alike models;
- Developed and implemented an algorithm predicting actual value of all client devices;
- Completed several geo-analytics projects;
- Worked on projects improving quality of customers’ service by call-center;
- Developed a prototype of human detection on video and predicting age/gender;

- Developed algorithm for predicting home and work station for users based on the logs of their activity.
- Developed algorithm for predicting user interests based on the visited sites.