Machine Learning Engineer

Machine Learning Engineer at AV Soft, Moscow (Okkruzhnaya, full-time office). Salary 150,000–250,000 RUB net. Develop ML models for information security products: anti-spam, deception, sandbox, multiscanner, TI.

AV Soft is a company in information security and data analytics for corporate clients. Products: anti-spam, deception, sandbox, multiscanner, TI. Responsibilities: - Develop and improve ML models: choose architectures, create features, train and validate models. - Collect and prepare data: parsing, cleaning, EDA, creating training samples. - Perform feature engineering and evaluate model quality by accuracy and performance metrics. - Conduct A/B testing and comparison of model versions. - Prepare models for operation: containerization in Docker and participation in deployment. - Test models, analyze false positives and customer cases. - Maintain technical documentation, describe architecture, features, experiment results, ensure model versioning. - Interact with developers, participate in code review and integration of models into products. - Monitor models in production: track quality and data drift, initiate retraining. - Support and develop ML pipelines and automation of model training and delivery.

Commercial experience in developing and bringing ML models to production from 1–2 years. Confident knowledge of Python, NumPy, Pandas, Scikit-learn, PyTorch, CatBoost, XGBoost, LightGBM, Transformers. Understanding of classical ML algorithms and principles of their application. Practical experience with SQL, Git, Docker, REST API. Experience with Airflow or similar tools for ML pipelines. Linux at confident user level. Understanding of MLOps principles, experiment management and versioning of models and data. Ability to write clean and maintainable code. English at the level of reading technical documentation. Nice to have: - Experience optimizing models with ONNX, TensorRT or similar tools. - Understanding of deep learning architectures, Attention and modern neural network models. - Knowledge of SHAP, LIME and methods for assessing vulnerabilities of ML models. - Experience integrating ML models into high-load services. - Experience with logs, network traffic, DNS, URL, HTML and understanding of basic network protocols and HTTP. - Experience in information security, anti-fraud, anti-spam or threat detection. - Relevant courses or certificates in ML/DL/MLOps. - Initiative and desire to develop products and propose new solutions.

Employment under the Labor Code of the Russian Federation, white salary (accredited IT company). Office within walking distance from Okkruzhnaya metro station. Training at the company's expense, corporate library, knowledge base, guides. Participation in complex projects in information security and data analytics.