Ml Engineer / Data Scientist (Deep Learning)

TL;DR

Ml Engineer / Data Scientist (Deep Learning): Building models to accurately predict future sales for tens of thousands of products in thousands of stores with an accent on optimizing inventory, planning purchases and managing promotions. Focus on deep learning models with PyTorch and/or TensorFlow for demand forecasting and time series tasks.

Локация: Москва (м. Добрынинская). Можно удалённо из РФ

Компания

X5 Tech is an accredited IT company and the main digital partner of X5 Group's retail chains and businesses, developing solutions to enhance technological comfort for employees and improve the shopping experience for customers.

Что делать

• Develop, train, and optimize deep learning models using PyTorch and/or TensorFlow for demand forecasting and time series tasks.
• Set up and maintain ML experiments, logging metrics, parameters, and model artifacts using MLflow or similar tools.
• Apply and develop model interpretability methods, including feature importance analysis and explainability models.
• Develop neural architectures for time series forecasting (LSTM, GRU, Transformer, TFT, etc.).
• Develop and improve data preparation pipelines for time series (feature engineering, lag generation, working with external factors, handling gaps and anomalies).
• Participate in the implementation of models in production, monitoring quality degradation and retraining models.

Требования

• Practical experience in developing deep learning models using PyTorch and/or TensorFlow.
• Experience with experiment tracking and model lifecycle management tools (MLflow or similar solutions).
• Ability to analyze and interpret models, including the application of feature importance assessment methods and explainability (SHAP, permutation importance, attention-based methods, etc.).
• Basic understanding of neural network architectures for time series forecasting (RNN, LSTM, GRU, Transformer approaches, TFT, etc.).
• Understanding of the principles of data preparation for time series (feature engineering, lags, exogenous features, handling gaps, normalization).
• Experience in assessing the quality of time series models and selecting metrics (MAE, RMSE, MAPE, WAPE, etc.).

Культура и преимущества

• Ensuring the stability and availability of the company's IT systems.
• Developing internal cloud services and digital projects.
• Acting as an IT business partner of the company and conducting business discovery.
• Turning X5 into a data-driven company and initiating new developments.