Разработчик систем искусственного интеллекта / ML engineer

About the role

We are looking for a Machine Learning Engineer specialized in time‑series forecasting to work on a 3‑month contract project. The candidate will design, implement and maintain AI‑driven prediction systems using Python and modern ML libraries.

Key responsibilities

• Develop and fine‑tune classical time‑series models (SARIMA, ETS, Prophet).
• Build gradient‑boosting solutions (XGBoost, LightGBM) with engineered lag and seasonal features.
• Design, train and evaluate deep learning models, including CNNs and transformer architectures.
• Implement walk‑forward validation and monitor custom business metrics (MAE, RMSE, MAPE, sMAPE).
• Detect and handle anomalies, missing values, data leakage and data drift.
• Process large datasets using ClickHouse, PySpark or Dask.
• Document experiments in reproducible Jupyter notebooks and versioned scripts.
• Package models with Docker and orchestrate pipelines with Airflow or Prefect.
• Participate in code testing and produce accompanying documentation.

Required profile

• Strong Python programming skills (pandas, numpy, scikit‑learn).
• Proficient SQL knowledge, including window functions, joins and aggregations.
• Hands‑on experience with time‑series forecasting methods and gradient‑boosting frameworks.
• Familiarity with convolutional networks and transformer models.
• Solid understanding of validation techniques for temporal data and relevant statistical tests.
• Experience with big‑data tools and cloud‑native deployment.

Required skills

• Python
• pandas
• numpy
• scikit‑learn
• SQL
• SARIMA
• ETS (Exponential Smoothing)
• Prophet
• XGBoost
• LightGBM
• Convolutional Neural Networks
• Transformers
• Walk‑Forward Validation
• MAE, RMSE, MAPE, sMAPE
• ACF/PACF analysis
• Dickey‑Fuller test
• ClickHouse
• PySpark
• Dask
• Jupyter Notebook
• Docker
• Airflow
• Prefect

What we offer

• 3‑month contract (contract‑based employment).
• Stable remuneration, amount discussed after interview.