Ver o conteúdo principal

Summer Course: MLOps

1920X670 Slide

Summer Course: MLOps

The summer course in "MLOps" will be held from July 11th to 15th, 2022 (from 2 p.m. to 7 p.m.), in person.

This course is a practical approach to modern software development (focused on agility, customer satisfaction, and repeatability) and machine learning topics, trying to shine a software engineering focus on the solution of problems using machine learning, in such a way that the final solution can be easily deployed in current cloud environments. It focuses not on specific tools, but on giving the student elements of the state of the art so that they can take their own decisions depending on the specific problem. The course will be entirely practical, and by the end of the course students will be able to create reproducible data science pipelines as well as write reports that will pass the publication filter.

This course will be taught in English.

Lecturer

Juan-Julián Merelo Guervós

Image content

JJ Merelo is professor and researcher at the university of Granada, and senior software engineer at polypoly. Almost 40 years continued developers experience, starting with the now retro ZX Spectrum. Published and released as open source several evolutionary algorithms and neural net libraries in Perl, JavaScript and Objective C. You can check his work here.

General Information

Dates to Remember

  • Period of attendance: from July 11th to 15th, 2022 (from 2 p.m. to 7 p.m.);
  • Payment: until July 7th, 2022;
  • The registration will be held until July 6th, 2022.

Length and ECTS

This course will last 25 hours and it confers 5 ECTS to participants who perform the assessment (project). This course will be taught in English.

Course Fee

The fee of the course is €450. There are special dicounts to Teaching staff, Students, Alumni and technical staff of Universidade Nova de Lisboa. For more information, please contact us to masters@novaims.unl.pt.

Teaching Methodologies (including evaluation)

We are going to use project-based learning (PBL), implying that the student has to engage on a project from the beginning, a project that has been chosen by them, and work on in in a small team to take it to completion by the end of the period. The projects will have to be (largely) automatically checked for correctness, with the teacher acting as product manager coordinating the efforts of the team and reviewing where needed.

The project will have to be managed with state of the art tools for software development, such as git and GitHub and their accompanying tools. The emphasis will be not so much on a bombardment of concepts, but on interiorizing a process to develop quality ML tools and deploy them continuously to where they’re needed (either in an industrial or an academic environment).

Evaluation Method

It will be continuous, student-centered evaluation. Although eventually the student will be required to produce a ML “artifact” (that includes the ML pipeline as well as an updatable system), the main evaluation will be how the student reaches the objectives that the rest of the team have created and how it responds to them.

Demonstration of the coherence between the teaching methodologies and the learning outcomes 

The methodology is based in a project, and the different learning outcomes are linked to this project. The different objectives are presented by the teacher in turn, and they will represent different milestones in the advancement of the project.