Working with data
You will be able to prepare and explore data, identify features, and notice problems that affect learning.
Master the fundamentals of machine learning—from data preparation and algorithm selection to model training, evaluation, and deployment. This course will explain how computer programs find patterns in data, make predictions, and help solve practical problems, such as classifying objects, estimating probabilities, grouping similar records, and analyzing images or text.
Theory will be combined with practical application in Python. You'll learn how to collect and explore data, identify gaps and errors, prepare features, divide the dataset into training and testing, select metrics, and compare model results. This workflow helps you understand why a model is making mistakes and what changes actually improve its performance.
You'll gradually move from classical algorithms to neural networks. You'll explore supervised and unsupervised learning, become familiar with reinforcement learning and transfer learning, and explore architectures for working with tables, images, sequences, and text. Special attention is paid to how to apply existing models and adapt them to specific tasks.
The program introduces the tools used in modern ML workflows: from data analysis libraries and training frameworks to model launch and integration tools. You'll learn how to select the appropriate tool, conduct experiments, and formalize the results into a project for further development.
This course is suitable for those seeking fundamental knowledge and a practical start in Machine Learning Engineering. You'll master the basics of working with ML models and create your own learning projects. The depth of individual topics and the range of technologies used will depend on the level of the group, the tasks, and the available computing resources.
Get a systematic introduction to machine learning: from data and basic algorithms to neural networks and performance evaluation.
Expand your development skills and learn to incorporate data processing, model training, and forecasting into software projects.
You'll understand the model-building workflow and be able to test hypotheses with data more confidently.
Understand how ML is used for forecasting, classification, pattern detection, and process optimization.
Consolidate your theoretical knowledge with practical experience using Python and machine learning libraries.
You will build a foundation for further study of neural networks, machine learning models, and their application in products.
The training is online and combines explanations of key concepts with hands-on Python work. You'll work through examples, run algorithms on data, compare predictions, and draw conclusions. As the course progresses, the tasks become more challenging, from analyzing an existing dataset to independently building and testing a model.
The practical training is built around the full ML cycle: formulate the problem, prepare the data, choose an approach, train the model, evaluate its quality, and explain the results. In the final project, you'll apply several of the learned steps to a single problem and present how the resulting model can be used further.
The internship will focus on Python. NumPy, Pandas, and SciPy will be used for data manipulation and computation, while Matplotlib and Seaborn will be used for visual analysis. You will use Jupyter Notebook or a similar environment to explore datasets, conduct experiments, and document results.
In addition to libraries, you'll learn the basics of working with GPUs and cloud computing, depending on your needs. It's important to understand the purpose of each tool: some help analyze data, others train models, and still others speed up or simplify their launch. Not all of the technologies listed are used in every project—the stack is selected based on the learning task and available hardware.
You'll start with basic neural networks and explore layers, activation functions, loss functions, gradient descent, and the main stages of training.
Explore how convolutional layers extract features from images and other spatial data. Explore image classification tasks and the typical workflow for preparing such data.
You'll get acquainted with models for sequential and time series data. You'll understand how LSTM and GRU models help take context into account at sequential steps, and compare them with modern approaches.
Learn the basic idea of the attention mechanism and the application of transformers to text and other sequences. You'll also gain practical experience using pre-trained models and adapting them to the task.
Explore the operation of a generator and discriminator that are trained in an adversarial process, and explore data generation learning problems.
The program provides a step-by-step introduction to the fundamentals of machine learning, data analysis, classical algorithms, and neural networks. Practical exercises will help you navigate the key stages of model building and assemble the final project.
You'll understand the problems machine learning solves, how data is converted into a forecast, and the stages of an ML project. Learn how to formulate a problem and select a measurable outcome.
Master the Jupyter workspace and get acquainted with Python tools for computing and data processing. Practice loading datasets and performing initial analysis.
Learn how to check data quality and handle missing values, outliers, and categorical features. You'll also explore sample splitting, normalization, and the risk of data leakage between training and validation.
Explore basic regression and classification models, learn how to build a simple ML pipeline, train an algorithm, and generate predictions using new examples.
You'll learn about metrics, validation, retraining, and parameter selection. Learn how to compare models and explain which solution is best suited for a given task.
Explore clustering, dimensionality reduction, and finding structure in data without ready-made answers. You'll try to interpret the resulting groups and the limitations of these methods.
You'll get acquainted with XGBoost, LightGBM, and CatBoost using case studies. Compare boosting results with other algorithms and explore the specifics of working with tabular data.
You'll explore the basic components of a neural network, the training cycle, and working with PyTorch and TensorFlow/Keras 3. You'll create a basic model and observe how its quality changes as you adjust the training settings.
You'll become familiar with key architectures for images, sequences, and text, as well as the generative adversarial approach. Compare their application areas and practice using existing models where appropriate.
Adapt a pre-trained model or refine the selected algorithm. Prepare the final project, evaluate its results, and consider integration or model execution options using ONNX Runtime, vLLM, or another suitable tool.
You will gain a holistic understanding of working on an ML task: from analyzing source data to selecting a model, checking its quality, and presenting the result.
You will be able to prepare and explore data, identify features, and notice problems that affect learning.
Learn to train basic classification and regression models, compare models, and select metrics.
Gain fundamental understanding of CNNs, RNNs, LSTMs, GRUs, Transformers, and GANs.
Explore the ecosystem of libraries for data analysis, classical ML, and deep learning.
Master Python tools for analysis and machine learning, and gain hands-on experience with classical algorithms and neural networks.
Understand transfer learning and get acquainted with pre-trained models from the Hugging Face ecosystem.
You will go through the key stages of the ML process and prepare a model for demonstration or further development.
$350 / month
Payment is by monthly subscription. The price includes two classes per week, practical assignments, and work on personal projects within the course curriculum.
You'll learn the fundamentals of working with data, classic machine learning algorithms, model training methods, and the main types of neural networks. The practical course will be conducted in Python.
The course introduces Python tools for data analysis. Basic Python and math skills will help you get started quickly; the depth of the requirements depends on the class size and format.
Supervised and unsupervised learning, reinforcement learning fundamentals, RLHF as a separate topic, and transfer learning.
Fully connected networks, CNNs, recurrent networks RNNs including LSTMs and GRUs, transformers and GANs.
The program includes Python data analysis tools, Scikit-learn, PyTorch, TensorFlow/Keras 3, MLX, Hugging Face Transformers, XGBoost, LightGBM, CatBoost, PyTorch Lightning, ONNX Runtime, vLLM, LangChain, and LlamaIndex. The specific toolset depends on the training tasks and hardware.
You'll learn the basics of neural network training and complete practical tasks using accessible examples. For some tasks, you'll also consider transfer learning and adapting pre-trained models.
A standard computer is sufficient for basic data processing and small modeling. Cloud computing or pre-built training environments can be used for more complex experiments.
The course will provide fundamental knowledge and practical skills for further development as a Machine Learning Engineer. Independent professional work typically requires continued practice and gaining experience through projects.
In classical reinforcement learning, an agent receives reward signals for its actions in the environment. RLHF uses human feedback and preferences to tune the behavior of the generative model; this is a separate introductory topic in the program.
You'll become familiar with transformers, existing models, and the principles of their adaptation. Full-scale training of large models from scratch requires significant computing resources, so the practical training will focus on training tasks and available methods for working with models.
You will present an ML training project, including problem definition, data preparation, a trained model, and evaluation of the results. The format and complexity of the project will depend on the chosen topic and the level of the group.