Trading Company

Developing high-frequency trading strategies. - Developed the feature extraction and feature selection pipeline.

Neurus

Developing CV models for production and research uses in various fields - face recognition, image classification, image generation, and character recognition. Developing and maintaining auxiliary software for data collecting and model performance monitoring. - Improved face recognition accuracy from ~94% to 99% by altering the embeddings postprocessing and by using the SOTA embeddings handling framework. - Improved user experience by accelerating SQL procedures during face recognition (the delay became imperceptible for a user). - Implemented automatic collecting of a dataset using the Telegram bot.

Market Research

This is the project I am developing myself. Developing Data pipelines (feature extraction, feature selection). Developing DL models (designing, training, validation). Developing trading strategies. - Developed a flexible research framework which facilitates conducting experiments on data and models. - Accelerated computing correlation by 4 orders. - Gained basic market predictability which is not a common pitfall and seems to be real.