Arthur Shamray
Data Scientist/ML engineer
Data science/Machine learning portfolio
Technical skills
- General: Python, Scikit-learn, PyTorch, Torchvision, NumPy, Pandas, Seaborn, Matplotlib, Requests, BeautifulSoup, SQL, JSON, Git
- Additional: Microsoft Excel, Exploratory data analysis, Microsoft Power BI
Education
-
B.S., Computer system analysis Dnipro University of Technology (Sep 2022 - Present) -
Associate’s degree, Computer engineering Dnipro Professional College of Energy and Information Technologies (Sep 2020 - Jun 2022)
Additional courses and trainings
- Machine learning at Hillel IT school (Jul 2023 - Sep 2023), Credential
Projects
-
Machine learning
- Delivery Duration Prediction: Tuned machine learning algorithm that predicts delivery duration from the Door Dash delivery company to the customer in seconds by some particular set of features. The dataset used in project is a public data posted by Door Dash company on the Strata Scratch website.
- Verification of ensemble models: This project was created to understand the differences between all available types of ensemble models, to check out their ROC AUC score and build confusion matrices. The dataset used in this project is public data that can be accessed via the link.
-
Deep learning
- CNN for motorcycles classifier: Convolutional neural network for predicting the class of motorcycle by image. Models were trained to predict three classes of motorcycles (sporbikes, city bikes and toutist bikes). The dataset consists of pictures of different classes of motorcycles collected from various search engines. The demo is available by the link.
- LSTM Neural Network for stock prices prediction: The main goal of this task is to create a model that will be able to make stock predictions based on previous values, general trends and other features; to test different algorithms and choose the best ones for each particular situation.
-
Data Analysis and visualization
- Exploratory Data Analysis (EDA) in Spotify songs dataset.: The purpose of this task is to give an overview on the open Spotify Songs dataset from Kaggle. To check what factors are more important in defining whether the song is going to be popular.
Contact me
If you enjoyed what you saw, want to have a chat with me, feel free to contact me on: