area , Master's degree — Generative AI Engineering
Coordinator: Michał Majewski, Ph.D. , Eng.
Symbol:
Specialized topics:
This specialization focuses on the design, training, optimization, and implementation of state-of-the-art generative and multimodal systems. The program covers the entire lifecycle of artificial intelligence systems—from the implementation of deep learning models, through their fine-tuning and evaluation, to production deployments and security considerations.
Students gain knowledge in the following areas:
- deep learning (CNNs, RNNs, autoencoders, GANs, Transformers),
- generative and multimodal models (Vision Transformers, diffusion models, CLIP),
- Retrieval-Augmented Generation (RAG) systems,
- effective model fine-tuning (LoRA, adapters),
- model compression and optimization (quantization, distillation, pruning),
- agent-based systems and model alignment (RLHF, DPO),
- reinforcement learning (including applications in robotics),
- AI and MLOps systems engineering (API, Docker, CI/CD, MLflow, Kubernetes).
The program combines theoretical foundations with intensive engineering practice, preparing students to work with modern foundation models and large-scale systems.
Ways to apply the knowledge gained in future professional work:
Graduates are prepared to work as AI Engineers, ML Engineers, or generative systems specialists. The skills they acquire enable them to design and implement systems based on generative models in industries such as IT, robotics, the creative sector, medicine, and data analysis. The program emphasizes practical implementation skills (MLOps, APIs, containerization), enabling graduates to work in both product and research and development teams.
Sample thesis topics:
- The Retrieval-Augmented Generation System for Knowledge Retrieval and Generation
- Multimodal image-text search system (CLIP)
- Optimization of LLM models (LoRA, quantization)
- An agent-based system using generative models
- Generative AI in Robotics (Perception, Planning, Control)
- Security and Robustness Analysis of Generative Models
(based on projects and course content)
Laboratory description:
The labs are project-based and engineering-focused. Students implement models from scratch (NumPy, PyTorch, TensorFlow), build generative systems (RAG, multimodal search, diffusion models), and then transform them into working services. These include API design (FastAPI), containerization (Docker), data pipelines, monitoring, and deployments in multi-service environments and Kubernetes. The classes develop the ability to work with real-world system constraints and computational costs.
Specialization courses:
Deep Learning in Python – fundamentals and advanced deep learning techniques, model implementation, and projects (CNN, RL, RAG).
Generative AI – modern generative and multimodal models, optimization, alignment, agent-based systems, and implementations.
Project-based classes:
PRO1 – AI and MLOps systems engineering: APIs, Docker, CI/CD, MLflow, Kubernetes, model-as-a-service deployment.
PRO2 – Trends in Generative AI: Scaling Laws, Security, Model Economics, and Future Directions.
PRO3 – Thesis Preparation: Academic Writing, LaTeX/Typst, Presentation of Results, Preparation for the Defense.
In addition, it is recommended that students take the elective course "Methods of Knowledge Engineering" in the 6th semester, which supplements their knowledge of expert models, machine learning, and knowledge engineering.
The choice of specialization at the Faculty of Computer Science takes place after completing the survey.
The survey is completed by second-year Master's degree students at the Faculty of Computer Science.


