In 2007, when I got my degree in Applied Mathematics from Moscow Institute of Physics and Technology (MIPT), artificial intelligence was still more of a concept than a working industry. It took over a decade of managing software development projects in finance, e-commerce and online gaming before I discovered my passion for data science and building machine learning systems with my own hands.
- Tools: Python (numpy, pandas, sklearn, matplotlib), R (data.table, ggplot2, caret), SQL
- Algorithms: prediction and classification, SVM, KNN, PCA, random forest, gradient boosting, ensembling, time series analysis, recommender systems, anomaly detection
- Deep learning
- CNN: image classification, object localization, object detection, landmark finding, face recognition
- RNN: GRU, LSTM, word embeddings, attention models, trigger word detection
- Technology: TensorFlow, Keras
- Statistical inference: hypothesis testing, regression models
Some projects that I’ve done:
WordAlign

Word alignment for Russian-Chinese bitext using BERT contextualized embeddings and further fine-tuning on unique parallel corpus
SwiftPredict

Ngram text prediction model for mobile keyboards that can be trained from scratch on a laptop, occupies only 219 MB RAM and provides suggestions within 22 msec
Ships detection in satellite imagery

96.77% F1 score detecting ships in satellite imagery with imbalanced classes, using transfer learning with VGG16 and Inception v3 CNNs
Formal education:
- Specialization in Data Science, Johns Hopkins University, 2021 (expected)
- Specialization in Deep Learning, deeplearning.ai, 2020
- Bachelor’s degree in Applied Mathematics, Moscow Institute of Physics and Technology (MIPT), 2007