DS412
Development of Modern MLOps Platforms
Faculty Profiles
Yevgeniy Ilyin
Senior Solutions Architect at Databricks
Nikita Fedkin
Senior Solutions Architect at Amazon Web Services (AWS)
Course length
Duration
Total hours
Credits
Language
Course type
Fee for single course
Fee for degree students
Skills you’ll learn
Overview
This three-week hands-on course provides a jumpstart into design and development of modern MLOps platforms. From initial project scoping to exploratory analysis and data engineering, from feature engineering to model training, to ML pipelines and CI/CD automation, productization, monitoring and operations.
You are going to create an ML project in Amazon SageMaker Studio and go through all stages of development such as data exploration, interactive experimentation, setting up MLOps pipelines, and finally deliver the project into production. You’ll learn how to work with a feature store, model registry, pipelines, model and data monitor, and CI/CD projects.
Through this course, you will work with Amazon SageMaker to get you inspired to implement your own MLOps solution. The course provides recommended patterns and practical architecture blueprints for real-world ML projects. The course gives you an overview of the industry-standard ML platforms and MLOps products, like MLflow, Apache Airflow, H20, DataIKU.
As the course main deliverable, you’re required to implement an end-to-end ML project with the main components of MLOps, such as reproducible ML pipelines, scalable data processing and model training, model registry, experiment tracking, observability, and event-driven workflows.
Learning highlights
- Understand practical concepts, architectural blueprints, and state-of-the-art patterns of MLOps.
- Gain hands-on experience by working on a real-world ML project.
- Learn industry relevant use cases and solutions.
- Gain essential working experience with AWS and Amazon SageMaker.
Course outline
15 classes
Session 1
ML refresher: Models, data preparation and processing, feature engineering, model training and testing, deployment and serving, Inference.
MLOps overview: What, why, and how MLOps. Main components and patterns of MLOps platforms.
ML pipeline: ML pipeline for model training, evaluation, and deployment. Practical vendor agnostic ML pipeline architectures.
Session 2
Experiments and model registry: Experiment tracking and model management at scale. MLflow integration.
Feature stores: Feature management, ingestion, and extractions. Online and offline feature stores. Feature store architectural patterns. Overview of feature store solutions.
CI/CD for ML: Automation and reproducibility for ML workflows. Applying the software development practices to ML.
Session 3
Introduction to AWS and Amazon SageMaker: SageMaker Studio and Studio notebooks, SageMaker processing and training jobs, deployment, hosting, and inference.
ML development with no-code AutoML: Amazon SageMaker Canvas and auto ML. How to implement an ML project end-to-end without writing a single line of code.
Session 4
Onboarding to Amazon SageMaker: SageMaker Studio and user profile setup, start Jupyter notebooks.
ML development with Amazon SageMaker: Create your first ML project. Implement an ML workflow: data processing, model training, deployment, and inference.
Session 5
Data and infrastructure: Compute infrastructure, containers, container image development, data storage and management, data lakes, data processing at scale.
Session 6
Generative AI: Foundational models, Large Language Models, use cases, design patterns.
Introduction into FMOps/LLMOps: Operationalization of generative AI applications, model customization and evaluation.
Session 7
Project development.
Guest lecture: refact.ai
Session 8
Project development.
Session 9
Project development.
Session 10
Self-paced work on the project.
Office hours: 11:00-12:00
Session 11
Operationalization: Observability, data and model monitoring, data and model quality, performance testing, security.
Responsible AI: Explainability, bias and fairness, bias mitigation throughout the ML model lifecycle.
Session 12
Project development.
Session 13
Project development.
Visit AWS office.
Session 14
Project development.
Session 15
Final project presentations - graded: Each team or individual presents own project, 30 min per team: 15 min demo + 15 min Q&A.
Course materials
Books
Media
Prerequisites
Students will need knowledge of basic Python programming and ML foundation, such as common models, basic ML development process, quantitative metrics, and inference. Basic foundational statistics and math are required. Basic understanding or hands-on experience of modern application development such as microservices, serverless, event-driven architectures, containers, and devops.
No previous AWS or Amazon SageMaker knowledge or experience is required.
Methodology
We expect students to spend approximately 20-30 hours on the course project. Project work is self-guided. We offer on demand office hours for project work support and questions.Students are free to choose any ML use case and develop it using MLOps design principles. The main motivation for the project is to demonstrate how to stage an ML solution from experimentation to production.
Project completion milestones:
Week 1: M1 - Project definition, scope, and design - graded. Hand-in by Friday June 14th 20:00 CET.
Week 2: M2 - Project MVP - graded. Hand-in by Friday June 21th 20:00 CET.
Week 3: M3 -Project demo - graded. Live demo on Friday June 28th
Grading
Yevgeniy Ilyin is a Sr. Solutions Architect at Databricks in Zurich. He received his master degree in mathematics at Moscow Institute of Physics and Technology and graduated with a Certificate Programme in Computer Science at the Swiss Federal Institute of Technology ETH Zurich. Yevgeniy is also a Chartered Financial Analyst (CFA) charterholder.
He has collected over 20 years of end-to-end experience working in the Financial Services Industry (FSI) in different verticals, such as asset and fund management, trading systems and order management, core banking and front end.
See full profileNikita Fedkin is a Solution Architect at Amazon Web Services (AWS) in Munich. He graduated from the Russian State University of Oil and Gas with a master's degree in applied mathematics. Since his career began, Nikita utilised his academic knowledge to solve cutting-edge business problems.
He started as a Data Science developer more than 10 years ago. Afterwards, he moved to the compute infrastructure field, became a System Architect of distributed payment gateway system, and finally became Head of Infrastructure of an International Auction house. Now he shares his knowledge with the world as a Solutions Architect at Amazon Web Services.
See full profileApply for this course
Development of Modern MLOps Platforms
by Yevgeniy Ilyin, Nikita Fedkin
Total hours
45 Hours
Dates
Jun 10 - Jun 28, 2024
Fee for single course
€1500
Fee for degree students
€750
How to secure your spot
Complete the form below to kickstart your application
Schedule your Harbour.Space interview
If successful, get ready to join us on campus
FAQ
Will I receive a certificate after completion?
Yes. Upon completion of the course, you will receive a certificate signed by the director of the program your course belonged to.
Do I need a visa?
This depends on your case. Please check with the Spanish or Thai consulate in your country of residence about visa requirements. We will do our part to provide you with the necessary documents, such as the Certificate of Enrollment.
Can I get a discount?
Yes. The easiest way to enroll in a course at a discounted price is to register for multiple courses. Registering for multiple courses will reduce the cost per individual course. Please ask the Admissions Office for more information about the other kinds of discounts we offer and what you can do to receive one.