Course for Collegium da Vinci University: Advanced Artificial Intelligence
I had the opportunity to prepare and deliver the Advanced Artificial Intelligence course for Collegium da Vinci University. The course was designed as a practical and structured introduction to modern AI, combining theoretical foundations with hands-on work in Python, machine learning workflows, neural networks, and Large Language Models. It also included AWS Academy materials, interactive labs, and project-based assignments to help students move from concepts to real implementation.
The program covered the broader landscape of artificial intelligence and machine learning, including the relationship between AI, machine learning, deep learning, and LLMs, as well as real-world use cases and examples of how these technologies are applied in practice. Students were introduced to different learning paradigms such as supervised learning, unsupervised learning, self-supervised learning, transfer learning, and reinforcement learning.
A significant part of the course focused on supervised machine learning algorithms, including linear regression, logistic regression, k-nearest neighbours (KNN), Naive Bayes, decision trees, random forests, ensemble methods, and Support Vector Machines. These topics were supported with activities and practical exercises to help students understand not only how the algorithms work, but also when they should be applied.
Major block was dedicated to artificial neural networks. Students explored the history and foundations of neural networks, from the McCulloch-Pitts neuron and the perceptron to shallow and deep neural networks. The practical part included understanding forward propagation, activation functions, gradients, overfitting and underfitting, regularization, learning rate, and model capacity. Interactive tools such as TensorFlow Playground and Python notebooks were used to make these concepts easier to understand in practice.
The course also addressed the machine learning process end to end: problem formulation, data collection, data analysis, feature engineering, data cleaning, handling missing values and outliers, model training, evaluation, deployment, and iterative improvement. Students worked with project templates and learned how to move from a business problem to a trained and evaluated ML model.
A dedicated part of the program focused on Large Language Models, including what text models are, how LLMs differ from traditional ML models, how transformers changed NLP, and how concepts such as context windows, tokens, model capacity, and training scale influence model behavior. This section connected foundational ML knowledge with the current generation of generative AI systems.
The course was strongly practical. Students were expected to build their own projects in small groups, using real datasets and implementing full ML workflows: preprocessing, feature engineering, model training, evaluation, tuning, and presentation of results. The evaluation criteria emphasized not only a working solution, but also code quality, use of version control, exploratory data analysis, testing, and interpretation of model metrics.
This course was an opportunity to combine my experience in machine learning, LLMs, software engineering, and technical education into a learning path that helps students understand both the foundations and the practical application of modern AI systems.