This course provides a practical, engineering-oriented introduction to Generative Artificial Intelligence, with an emphasis on modern, industry-relevant techniques rather than classical generative models. Students learn how large language models, diffusion models, and multimodal systems work, as well as how to deploy them locally on GPUs to build real-world applications for text, image, audio, and code generation. Through hands-on laboratories, the course covers prompt engineering, agents, retrieval-augmented generation (RAG) systems, lightweight fine-tuning, and performance optimization, while also addressing key considerations regarding safety, bias, and regulation. By the end of the course, students will be able to design, implement, and defend a complete GenAI application following professional best practices.
Professors
There are no prerequisites.
By the end of this course, students will be able to:
- Understand the operating principles of the main modern Generative Artificial Intelligence systems, including large language models (LLMs), diffusion models, and multimodal systems.
- Understand the role of transformer architectures, latent spaces, and foundational models in the recent evolution of GenAI.
- Deploy and run generative models locally on GPUs, efficiently managing resources such as VRAM, quantization, and inference performance.
- Design and implement complete pipelines for generating text, images, audio, voice, and code.
- Build agent-based systems, use tools, and implement retrieval-augmented generation (RAG) using vector stores.
- Apply lightweight fine-tuning techniques to adapt models to specific tasks.
- Integrate GenAI models into functional applications such as web services, command-line tools, or intelligent assistants, following software engineering best practices.
- Develop technical judgment to make and justify decisions regarding architecture, performance, and deployment.
- Identify and analyze the limitations of generative systems, such as hallucinations, biases, security vulnerabilities, and misuse risks.
- Understand the ethical and regulatory framework applicable to GenAI, with special attention to the European Artificial Intelligence Act (EU AI Act).
- Promote teamwork, technical communication, proper documentation, and the public defense of engineering solutions.
1.Introduction to Generative AI (Modern View)
2.Light Foundations: Latent Spaces & Classical Generative Concepts
3.Transformers and Sequence Models
4.GPT architecture, scaling laws, alignment
5.Building Applications with LLMs
6.Diffusion Models & Image Generation
7.Multimodal models
8.Audio, Speech & Music Generation
9.Code Generation Models
10.GenAI Safety, Bias & Regulation
Weekly teaching will consist of two lecturing sessions to explain basic concepts to apply knowledge to practical situations. Exercises in class will be solved and problems will also be proposed so that students can apply the concepts learned.
Each topic will have the following elements:
| Elements | Methodology |
| Concepts review | Lecture |
| Examples | Class review |
| Labs | Problem solving |
Continuous assessment has the following evaluation structure:
Grading:
- Individual assignments: 20%
- Group assignments: 20%
- Final group projects: 50%
- Presentation (group): 20%
- Written report (group): 20%
- Oral Q&A (individual): 10%
- Attendance, participation, and engagement in class: 10%
The evaluation criteria apply to all the students, retakers must attend to class and submit all the deliverables requested. Any exceptional situation should be communicated previously to the professors and validated by the tutor.
The subject will be passed if the overall score is higher than 5.
RETAKE POLICY
The Retake will consist of an exam that includes all the content of the subject.
All case studies homework must have been submitted to virtual campus before retake exam. Otherwise, the retake exam will not be possible.
Use of Generative Artificial Intelligence
This course adopts the AIAS (Artificial Intelligence Assessment Scale) to regulate the use of
generative AI in assessment activities:
- Level 4: The use of generative AI is permitted for specific parts of an assignment, provided that students critically evaluate the generated output and appropriately acknowledge and document its use.
Continuous assessment has the following evaluation structure:
Grading:
- Individual assignments: 20%
- Group assignments: 20%
- Final group projects: 50%
- Presentation (group): 20%
- Written report (group): 20%
- Oral Q&A (individual): 10%
- Attendance, participation, and engagement in class: 10%
The evaluation criteria apply to all the students, retakers must attend to class and submit all the deliverables requested. Any exceptional situation should be communicated previously to the professors and validated by the tutor.
The subject will be passed if the overall score is higher than 5.
RETAKE POLICY
The Retake will consist of an exam that includes all the content of the subject.
All case studies homework must have been submitted to virtual campus before retake exam. Otherwise, the retake exam will not be possible.
Use of Generative Artificial Intelligence
This course adopts the AIAS (Artificial Intelligence Assessment Scale) to regulate the use of
generative AI in assessment activities:
- Level 4: The use of generative AI is permitted for specific parts of an assignment, provided that students critically evaluate the generated output and appropriately acknowledge and document its use.
The bibliography and course materials will be reviewed and updated throughout the course in accordance with developments in the field and the topics covered.
The bibliography and course materials will be reviewed and updated throughout the course in accordance with developments in the field and the topics covered.