Vladislav Soldatov

regression, clustering (on tabular data), time series forecasting; Tech stack: python, statistics and machine learning libraries (statsmodels, scikit-learn, tensorflow/keras)

Warehouse optimization project to reduce labour costs; Time series analysis and forecasting (python (statsmodels, prophet, darts), PostgreSQL, Dash web apps)

Improved online ad targeting using age, gender and interest prediction; Apache Spark based machine learning pipeline (deployed on AWS) for web tracking data featurization, machine learning model training and prediction; Improved advertising segment reach by finding similar users (look-alike, unsupervised learning, high-dimensional similarity search)

Implemented a collaborative recommender system to increase user engagement with online content; Conducted a research review to find the most suitable algorithms (online learning, neighbourhood based model, approximate nearest neighbour search with Annoy)

Qt-based collaborative editing components for geophysical data, research and building a prototype

Bayesian techniques for discrete stochastic volatility models (discrete stochastic volatility models augmented with trading volume, volatility forecasting using Markov Chain Monte Carlo for calibration, MATLAB)

Econometrics of cross-sectional, time series and panel data (teaching and research)

Wavelet analysis of mechanical vibrations (MATLAB)