МОВС MLOps (2023-24 уч. год, 2 модуль)
Содержание
О курсе
The MLOps course covers a wide range of topics essential for effectively managing and deploying machine learning projects. It begins with an introduction to MLOps, providing an overview of principles and best practices. Students gain hands-on experience in creating ML projects using Python and learn about containerization using Docker. Code management and version control using Git servers are emphasized, enabling collaborative development. The course explores Continuous Integration and Continuous Deployment (CI/CD) techniques for automating the ML workflow. Data management, including storage, versioning, and management using tools like DVC, is covered. Students learn about experiment logging to track and analyze model performance. The course addresses computational power requirements, cloud-based solutions, and transferring learning to the cloud. Additional topics include knn indices, embeddings quantification, crowdsourcing, data labeling using platforms like Toloka, and an overview of the Amazon SageMaker stack. Talks by students and external lecturers provide practical insights. By course completion, students gain a comprehensive understanding of MLOps principles and practical skills for managing and deploying ML projects.
Занятия проводятся в Zoom по вторникам в 17:30 и субботам в 10:30
Контакты
Чат курса в TG: https://t.me/+VBBJVl19s5BiYmU6
Преподаватель: Гончаренко Владислав, Dzen.ru
Ассистенты
Ассистент | Контакты |
---|---|
Владислав Наумов | @vlad21naumov |
Алексей Макарчук | @alexmak123 |
[1] |
Материалы курса
Плейлист курса на YouTube: https://www.youtube.com/playlist?list=PLmA-1xX7IuzCRcPf_aGos3d2uvAQrd46c
GitHub с материалами курса: GitHub repository
Записи консультаций
Формула оценивания
Оценка = Среднее за домашние задания
Домашние задания
- (1-й курс)
- (2-й курс)
- Poetry Setup
- Code Quality Tools Setup
- DVC Setup
- MLFlow Setup
- Running on Remote Machine
- ...
Полная формулировка задачи для проекта
Литература
- "Machine Learning Engineering" by Andriy Burkov, 2020
- "Practical MLOps: Operationalizing Machine Learning Models (2021)", Noah Gift & Alfredo Deza