Alina Khazhieva
Machine Learning Engineer at Quantum Brains- Claim this Profile
Click to upgrade to our gold package
for the full feature experience.
Topline Score
Bio
Experience
-
Quantum Brains
-
Germany
-
Research Services
-
1 - 100 Employee
-
Machine Learning Engineer
-
Jun 2023 - Present
Тбилиси, Грузия
-
-
-
InnoStage
-
Russian Federation
-
IT Services and IT Consulting
-
100 - 200 Employee
-
Machine Learning Engineer
-
Jun 2019 - Jan 2023
Иннополис Building a model that detects tunnelling in DNS using an ensemble of models of Random Forest and Multi-Layer Perceptron. It processed streaming data over 5 minutes. Detecting anomalies in Time-Series using LSTM. Full development of the antifraud system with machine learning using Python, including: - Studying state-of-the-art solutions for detecting fraudulent transactions; - Creating aggregated features using SQL queries and pandas library; - Making Exploratory Data… Show more Building a model that detects tunnelling in DNS using an ensemble of models of Random Forest and Multi-Layer Perceptron. It processed streaming data over 5 minutes. Detecting anomalies in Time-Series using LSTM. Full development of the antifraud system with machine learning using Python, including: - Studying state-of-the-art solutions for detecting fraudulent transactions; - Creating aggregated features using SQL queries and pandas library; - Making Exploratory Data Analysis (EDA) on the features with the scikit-learn library and visualisation of the results using seaborn and matplotlib libraries; - Preprocessing numerical and categorical features for future training using scikit-learn (ColumnTransformer, MinMaxScaler, SimpleImputer, LabelEncoder); - Training classification models, such as SVM, Logistic Regression, and CatBoost. All models were hyperparameter tuned with Random Search; - Feature selection using CatBoost library (select_features based on SHAP values); - Development a script for training pipeline with validation and saving the resulting model in the pickle file; - Development of a microservice on FastAPI for the classification of transactions in real-time and updating aggregated features by schedule. It included validation of input data with pydantic library. - Writing unit tests for the training script and predicting microservice using pytest library. Work on Airflow: - Transferring updates of aggregated features into DAGs with alerts if the code crashes; - Development of a fully automated re-training of the model and saving the model file into the S3 storage; - Writing unit tests and integration tests for DAGs. Show less
-
-
-
RockyLabs
-
Russian Federation
-
Software Development
-
1 - 100 Employee
-
Стажировка
-
Jun 2018 - Jul 2018
Казань, Россия Работала на должности full-stack developer, в больше части front-end. Разрабатывали платформу для бронирования столиков в ресторане. Были использованы такие стек-технологии: Vue.js, Typescript, MongoDB, Javascript, GraphQL
-
-
Education
-
Университет Иннополис