Data can be a valuable resource for companies, as it helps in decision making, performance analysis, and problem-solving. Estimates indicate that an average of 2.5 quintillion bytes of data are generated daily. The use of technology like Artificial Intelligence (AI) is indispensable for extracting useful information from this large amount of data. Companies that provide Cloud Computing services, such as Amazon with its Amazon Web Service (AWS) platform, allow the easy deployment of AI solutions capable of taking advantage of this immense volume of data. An example of these services is AWS Lambda, one of the main tools of what is known as serverless computing. Within this context, this mini-course will introduce the concepts of Function as a Service (FaaS), which encompass the use of AWS Lambda to provision AI applications.

Machine learning (ML) models and solutions have many applications in our daily lives, with the real possibility to improve products, processes, and techniques; however, in some situations, the systems do not sufficiently or convincingly explain their predictions. Understanding decision-making in highly sensitive areas such as healthcare or finance is of paramount importance to provide transparency, security, and trust to the users of intelligent systems. This short course introduces the participant, in theory and practice, to strategies and techniques for making machine learning models and their decisions interpretable and auditable. ResumoOs modelos e soluc ¸ões de aprendizado de máquina (ML) possuem diversas aplicac ¸ões no nosso cotidiano, com real possibilidade de melhorar produtos, processos e técnicas; entretanto, em algumas situac ¸ões, os sistemas não explicam de forma suficiente ou convincente as suas previsões. Entender a tomada de decisões em áreas altamente sensíveis, como saúde ou financ ¸as, é de suma importância para oferecer transparência, seguranc ¸a e confianc ¸a aos usuários dos sistemas inteligentes. Este minicurso apresenta ao participante, na teoria e prática, as estratégias e técnicas para tornar os modelos de aprendizado de máquina e suas decisões interpretáveis e auditáveis.