Digital Skills Accelerator - Machine Learning Operations Bootcamp

• I. General Information
• II. Calendar and key dates
• III. Communication Channels
• IV. Sessions
• V. Mentoring Model
• VI. Final Projects
• VII. Teams
• VIII. Program
• IX. Session Recordings and Slide Presentations
• X. Contact

📑 Index

• Module 1 - Introduction to ML and why MLOps
• Module 2 - Fundamentals of Mathematics and Computing for ML
• Module 3 - ML Systems Architecture
• Module 4 - ML Fundamentals
• Module 5 - Definition of business requirements
• Module 6 - Life Cycle of an ML System in Production
• Module 7 - Experiment administration, reproducibility, and traceability
• Module 8 - Model orchestration and model packaging
• Module 9 - Automation and monitoring

I. General Information

The Machine Learning Operations Bootcamp is carefully designed to help participants better understand the efficient management and deployment of machine learning models in real-world contexts. It covers topics ranging from the fundamental principles of machine learning to the intricacies of model deployment and monitoring.

In this training, participants will gain essential knowledge and skills for the smooth execution of machine-learning projects in production environments. Participants will integrate the expertise of four distinct technological profiles: Data Science, Data Engineering, Systems Reliability Engineering, and Software Engineering.

At the end of this learning journey, within their role, they will be ready to work collaboratively, optimize model performance, ensure data reliability, and maintain system efficiency throughout the project lifecycle.

II. Calendar and Key Dates

Link to Calendar

III. Communication Channels

These channels ensure effective and timely communication; use them as needed.

Wizeline Slack Channels Channels where general collaboration among Subject Matter Experts (SMEs) and specific roles is encouraged. These channels serve to stay informed about the progress and events of the training.

Wizeline Academy Slack Channels Channels for collaboration with participants.

• Slack MLOps Bootcamp: #mlops-bootcamp We will interact with the participants through this channel to share general announcements, provide technical support, and address general questions.

• Slack/Forum: #foro-mlops-bootcamp Channel dedicated to answering questions about the content. Further details are provided in Section V.

IV. Sessions

• All sessions are virtual - Zoom Link
• During the sessions, we recommend having your camera turned on and use the bootcamp's background.
• As the main instructor or mentor, you must attend a minimum of 5 minutes before each session.
• If you are unable to attend a session, please inform the management team and/or register your PTOs in the following PTO file.
• Encourage participants to contact us via email at:([email protected])
• Participant feedback is crucial for improvement. We kindly ask you to share the following session feedback form with your participants before each session comes to an end.

V. Mentoring Model

Mentorship sessions are designed to help participants address questions about the content and receive support for their projects. They are a valuable tool for receiving guidance from our experts in the four profiles.

The mentorship period for this bootcamp is from January 24 to May 17, 2024, and three approaches are being implemented.

1. Slack/Forum: #foro-mlops-bootcamp

• A channel dedicated to resolving doubts about the content.
• In this space, everyone can respond.
• The team of instructors and mentors must monitor and respond to questions within 24 hours.
• When sharing questions, it is suggested to encourage the use of the following structure: Topic: Indicate the topic of your question. Question: Formulate your question clearly and specifically to receive the best possible help. Example: Topic: Data Strategy Question: How is data strategy incorporated to handle the privacy and security of sensitive information during the machine learning lifecycle?

There are two types of mentorship sessions:

2. "Office Hours" Mentorship:

• Every Wednesday from 5 pm to 6 pm (Mexico City time) - Zoom Link
• Space where participants can ask our experts questions in the four technical profiles.
• If you are a mentor, check the office-hour mentoring calendarto confirm your attendance.

3. On-Demand 1:1 Mentorship:

• Participants can schedule a session by completing our mentorship request form.
• Based on their information, we will assign the most suitable mentor by consulting mentor availability via our mentor channel #dsa-coecytjal-mentors.
• Within 24 hours, participants will be notified and sent the link via Google Calendar.
• As a mentor, you'll need to provide feedback on your mentorship via our mentoring follow-up form. Your comments are crucial for training improvement.
• After your mentoring session, encourage participants to complete the following mentorship feedback form.

VI. Final Projects

The final projects of the training are divided into two essential parts to solidify participants' learning and practical application:

Technical Validation (Individual Project): Participants can individually apply the technical knowledge acquired during the training in this phase. They will be assigned a specific challenge they must complete as they progress through the topics.

• The information for this validation will be released to the participant starting from Module 3.

Business Validation (Group Project): This phase encourages collaboration and the practical application of skills in a group setting. Participants will work together to address a challenge. This group project will allow them to integrate their technical knowledge with a strategic and business-oriented vision.

• The information for this validation will be released starting from Module 5.

The purpose of these final projects is to consolidate participants understanding and skills, providing them with the practical experience necessary to tackle real-world challenges in the field of Machine Learning Operations.

VII. Teams

Team	Area	Names
Management Team	DSA	- Ana Paula Barragán - Violeta Baltazar - Fátima Pedroza
Technical Leadership Team	Data Science Data Engineering SRE SWE	- Ricardo Valdez - John Sánchez - Enrique Cuevas - Gerardo Ruiz

Team	Module	Names
Team of Instructors	Module 1	- Brenda Leyva - Roberto Galindo
Team of Instructors	Module 2	- Francisco Villalobos - Camilo Martínez
Team of Instructors	Module 3	- Grisell Reyes - Edgar Talledos
Team of Instructors	Module 4	- Roberto Galindo - Jorge Martínez
Team of Instructors	Module 5	- Daniela Villalobosa - Nicolás Losada
Team of Instructors	Module 6	- Grisell Reyes - Edgar Talledos
Team of Instructors	Module 7	- Luis Morales - Cristian Zapata
Team of Instructors	Module 8	- Brenda Leyva - Porfirio Hernández
Team of Instructors	Module 9	- Edgar Talledos - Goldy Dudhwa

Profile	Mentor
Data Science	- Porfirio Hernández - Luis Morales - Daniela Villalobos
Data Engineering	- Grisell Reyes - Francisco Villalobos - Martin Alarcón
SRE	- Edgar Talledos - Gonzalo Romero - Josúe Ruiz
SWE	- Gerardo Ruiz - Álvaro Rodríguez - Matías Ponce

VIII. Program

The program comprises nine modules delivered in 18 weeks, two weeks per module.

📑 Modules

Module 1 - Introduction to ML and why MLOps

Learner's Outcomes: Understand the fundamentals of machine learning. Recognize the importance of MLOps in real-world applications. Comprehend the significance of data strategy and governance in machine learning projects.

• AI introduction and Business Acumen
• Exploratory Analysis
• Data Strategy
• Data Governance
• What is MLOps and why do we need it

Module 2 - Fundamentals of Mathematics and Computing for ML

Learner's Outcomes: Develop a foundational knowledge of discrete mathematics. Acquire a working understanding of linear algebra. Learn essential probability and statistics concepts relevant to machine learning. Explore the theory of computation and its relevance in machine learning.

• Discrete Mathematics
• Linear Algebra
• Probability and Statistics
• Theory of Computation

Module 3 - ML Systems Architecture

Learner's Outcomes: Gain insight into the role of databases in machine learning systems. Understand the principles of handling big data in ML. Differentiate between various levels of ML system maturity. Learn the principles behind designing ML pipeline architectures. Distinguish the key differences between DevOps and MLOps in the context of ML systems.

• Databases
• Big Data
• ML System Maturity Levels
• ML Pipeline Architecture Design
• Differences between DevOps and MLOps

Modules 4 - ML Fundamentals

Learner's Outcomes: Identify different types of ML models and their use cases. Master the art of feature engineering and parameter optimization. Learn how to select and train a suitable ML model. Cultivate a mindset oriented towards MLOps in machine learning projects.

• Types of ML Models
• Feature Engineering and Parameter Optimization
• Selecting and Training a Model
• Adopting the MLOps mindset

Modules 5 - Definition of business requirements

Learner's Outcomes: Define key performance indicators (KPIs) and functional requirements for ML projects. Understand the concept of optimization and its relevance to ML projects. Recognize the types of use cases that can benefit from machine learning and their advantages.

• KPI definitions, functional requirements
• What is optimization?
• What types of use cases work to solve ML and its benefits?

Modules 6 - Life Cycle of an ML System in Production

Learner's Outcomes: Grasp different programming paradigms used in ML projects. Gain insight into the structure of an ML project. Explore the reasons for deploying ML models and pipelines in production environments.

• Programming Paradigms
• Structure of an ML project
• Why deploy ML models and pipelines?
• Different types of inferences and model requests

Modules 7 - Experiment administration, reproducibility, and traceability

Learner's Outcomes: Learn how to create reproducible ML pipelines. Identify and address challenges related to model reproducibility. Familiarize yourself with testing ML pipelines. Explore the use of tools like MLflow and DVC for experiment traceability.

• Creating Reproducible ML Pipelines
• Challenges to model reproducibility
• Testing ML pipelines
• Tools for traceability of experiments (MLflow/DVC)
• Traceability of experiments with (MLflow/DVC)

Modules 8 - Model orchestration and model packaging

Learner's Outcomes: Understand different methods of model deployment. Learn about patterns and deployment infrastructure in ML. Master the process of model registration. Explore options for model orchestration, including Kubernetes, ECS, Docker, and managed services. Learn to convert a notebook into a functional ML pipeline.

• Different methods of model deployment
• Patterns and deployment infrastructure
• Model registration
• Orchestration options (Kubernetes, ECS, Docker)
• Deploying via containers (managed services: sagemaker, AI factory, etc.)
• Converting a Notebook to a pipeline
• Orchestration of the ML pipeline

Modules 9 - Automation and monitoring

Learner's Outcomes: Gain an understanding of data and model drifting in ML systems. Learn how to monitor ML systems and web services effectively. Understand the importance of metrics in managing ML systems. Get introduced to continuous integration and deployment (CI/CD) in MLOps. Explore test automation with tools like GitHub Actions.

• Introduction to data/model drifting
• Monitoring ML systems
• Monitoring Web Services
• Metrics for ML Systems
• Intro to CI/CD
• Test automation with GitHub Actions

IX. Session Recordings and Slide Presentations

• Recordings
• Slide Presentations

X. Contact

If you have questions about the MLOps program, logistics, access, etc., please contact the Wizeline Academy DSA team.

📧 Email: MLOps Bootcamp ([email protected])