How to Learn MLOps in 2024 [Courses, Books, and Other Resources]

With the recent rapid advancements in machine learning and its widespread application in various industries, Machine Learning Operations (MLOps for short) has become a rapidly advancing field. The recent popularity of Large Language Models (LLMs) has only accelerated this development and increased the demand for MLOps specialists.

MLOps sits at the intersection of data science, DevOps, and software engineering and is all about reliably deploying and operating ML models at scale. In many ways, MLOps is an extension of DevOps. It comprises numerous activities:

• Data engineering and management
• Providing the infrastructure for model development and experiment tracking
• Creating and orchestrating training pipelines
• Managing data and model artifacts
• Deploying models to various cloud and hardware platforms
• Model and system monitoring

MLOps is a highly sought-after skill. But few data scientists, ML engineers, and software engineers have mastered it. Therefore, picking up some MLOps knowledge is a great career move in 2024.

Learning MLOps can be challenging because it is a vast and relatively nascent field. Tools and techniques are constantly evolving, with best practices just being discovered. If you’re just beginning your MLOps journey, it can be challenging to find accessible and up-to-date resources among all the marketing buzz.

If that’s you, you’ve come to the right place! For this article, we have sifted through a ton of resources and have picked out the ones we found to be of high quality. We’ve also compiled a learning roadmap to help you decide where to start.

Neptune’s MLOps learning roadmap

There is a lot to learn in MLOps. The best path depends on your background. If you come in from a software engineering or DevOps background, you’ll likely want to start with the operational aspects. As a data scientist, you’ll want to set off from familiar shores and work your way into MLOps from a model-centric perspective.

The roadmap below gives a general view of the skills and tools you need to know to truly master MLOps.

I encourage you to take some time to explore the roadmap and determine the best starting point for your MLOps journey. It’s also good to know your learning style and preferences when deciding between different options.

With that in mind, let’s explore the best resources for learning MLOps in 2024!

MLOps courses

If you’re just beginning with MLOps, a structured course is the best way to gain the foundational knowledge you need to navigate the space and solve problems.

I recommend the following courses taught by top industry experts:

1. Coursera’s MLOps Specialization by DeepLearning.AI
2. Stanford’s CS 329S: Machine Learning Systems Design
3. Made with ML’s MLOps Course
4. Full Stack Deep Learning
5. Vendor-specific MLOps courses by major cloud platforms

Jump right into the comparison of all MLOps courses.

1. Coursera’s MLOps Specialization by DeepLearning.AI

About: Andrew Ng’s MLOps specialization on Coursera is a four-course series that introduces developers to MLOps, specifically the TensorFlow ecosystem. It’s a top choice for MLOps education for $49 per course (you can also access it for free by auditing the course or applying for financial aid).

It teaches you the intricacies behind designing an end-to-end ML production system, like project scoping, data needs, modeling strategies, establishing a model baseline, and building data pipelines, all using tools from the TensorFlow Extended (TFX) library. It also highlights hardware requirements and how to implement distributed processing and parallelism techniques to maximize resources like compute, storage, and I/O resources in production environments.

Ideal for: Junior ML engineers and data scientists new to MLOps.

Curriculum scope: Here is an overview of the courses’ scope:

• Introduction to Machine Learning in Production
  • Overview of the ML Lifecycle and Deployment
  • Select and Train a Model
  • Data Definition and Baseline
• Machine Learning Data Lifecycle in Production
  • Collecting, Labeling and Validating Data
  • Feature Engineering, Transformation and Selection
  • Data Journey and Data Storage
  • Advanced Labeling, Augmentation and Data Preprocessing
• Machine Learning Modeling Pipelines in Production
  • Neural Architecture Search
  • Model Resource Management Techniques
  • High-Performance Modeling
  • Model Analysis
  • Interpretability
• Deploying Machine Learning Models in Production
  • Model Serving: Introduction
  • Model Serving: Patterns and Infrastructure
  • Model Management and Delivery
  • Model Monitoring and Logging

Key learning objectives: 

• Understand MLOps fundamentals.
• Learn how to build data and model pipelines in TensorFlow.
• Practice deploying and monitoring TensorFlow models in production.

Recommendation: I recommend this course because it gives you a solid foundation, even if you’re eventually working with a framework different from TensorFlow. Andrew Ng is well-known as an excellent educator, and this course does not fall short. If model interpretability is your top priority, you should check out this course.

Course reviews: The reviews for “Coursera’s MLOps Specialization by DeepLearning.AI” are overwhelmingly positive. Students praise the course for its practical approach and the clarity of the content. They highlight that Andrew Ng’s signature teaching style makes complex concepts easy to understand, and they appreciate the course’s focus on real-world examples. Students also note the emphasis on shifting from a code-centric to a data-centric mindset, a crucial aspect in the application of machine learning.

Here are some examples:

• Coursera Machine Learning Engineering for Production (MLOps) Specialization Review
• Is Coursera’s Machine Learning Engineering for Production (MLOps) Specialization Worth It in 2023? Review

2. Stanford’s CS 329S: Machine Learning Systems Design by Chip Huyen

About: Stanford’s CS 329S: Machine Learning Systems Design is an in-depth course taught by Chip Huyen in Stanford’s computer science curriculum. Huyen is well known for her high-quality blog articles and an MLOps book that originated in the lecture notes for a previous edition of CS 239S. She previously worked on deep learning systems at NVIDIA and on MLOps systems at Snorkel.

Compared with other courses in this article, CS329S focuses more on ML systems design than model building or other aspects of MLOPs. You will learn about data, infrastructure, hardware, and solution interfaces and how they all come together with the ML model to give you a solution in the real world. Some unique topics it covers include flow management using a scheduler, compiling and optimizing models for either edge devices or the cloud, container orchestration, evaluating MLOps tools for system compatibility, and data security. Data security, in particular, is not a topic you can find being discussed in most tutorials.

Ideal for: MLOps beginners with a basic understanding of machine learning. A background in Python programming is required for the practical exercises and projects included in the curriculum.

Curriculum scope: Here is what the course curriculum spans:

• ML and Data Systems Fundamentals
• Feature engineering
• Understanding machine learning production
• Model selection, development, training, and evaluation
• Diagnosis of ML system failures
• Monitoring and continual Learning
• Data distribution shifts on streaming data
• Experiment tracking and versioning with Weights & Biases
• Deploying time series forecasting and graph neural networks
• ML beyond accuracy: fairness, security, and governance
• ML infrastructure and platform
• Integrating ML into business

Key learning objectives:

• Foundations of machine learning and data systems, encompassing data management, feature engineering, and model selection.
• Considerations for and challenges of ML production deployments.
• Experiment tracking and model monitoring.

Recommendation: While not a typical online course for self-study, I recommend you consider going through the material because it is in-depth and teaches through practical examples. 

3. Made With ML’s MLOps Course

About: Made with ML’s MLOps course is a self-paced tutorial that shows you how to combine machine learning with software engineering to design, develop, deploy, and iterate on production ML applications. It also takes things further by covering LLMOps (MLOps for Large Language Models). The course is free and uses open-source tools and libraries.

MadeWithML regularly launches live cohorts, where you go through the course with others and can join live workshops and QA sessions. They also sometimes give out free GPU resource credits to developers. The course teaches you the intuition, tools, and code involved in the end-to-end development and deployment of ML systems. In this course, you will build and deploy a language model from end to end with MLOps and other general programming best practices in mind. This course is my personal favorite because it is a very hands-on resource.

Ideal for: Beginner to intermediate-level data scientists, ML engineers, and software developers.

Curriculum scope: Here’s what the scope spans for each component of the MLOps system design and implementation:

• Design
  • Setup
  • Product
  • Systems
• Data
  • Preparation
  • Exploration
  • Preprocessing
  • Distributed
• Model
  • Training
  • Tracking
  • Tuning
  • Evaluation
  • Serving
• Developing
  • Scripting
  • CLI
• Utilities
  • Logging
  • Documentation
  • Styling
  • Pre-commit
• Testing
  • Code
  • Data
  • Models
• Reproducibility
  • Versioning
• Production
  • Jobs and Services
  • CI/CD workflows
  • Monitoring
  • Data engineering

Key learning objectives: 

• Set up your environment for MLOps.
• Understand system design and how to build ML systems that scale. 
• Learn the foundational concepts of data and model pipelines.
• Practice migrating code from a notebook to Python scripts and utility files for testing, model versioning, and deployment.
• Learn CI/CD workflows, model monitoring, and data engineering concepts relevant to production ML.

Recommendation: This resource is a detailed yet concise blend of concepts and code. It will give you an understanding of both the theoretical and practical aspects of MLOps. If you prefer a self-paced course without sitting through video lectures, MadeWithML’s course is my top recommendation for you.

Course reviews: The reviews for the course are highly positive. Learners appreciate its practicality and immediate applicability to their jobs. The course is commended for providing insights into making optimal choices in ML engineering for various real-world use cases. It’s also noted for its approach to mimicking the production ML thought process, offering alternatives with different levels of complexity and weighing their pros and cons.

Here are some reviews of this course:

• Sherry Wang (Senior ML Engineer – Cars.com) – “Made with ML is one of the best courses I’ve ever taken. The material covered is very practical; I get to apply some of them to my job right away.”
• Lawrence Okegbemi (ML Engineer, Enterscale) -“Following all through, it’s really amazing to see how you demonstrated best practices in building an ML-driven application.”
• Dmitry Petrov (CEO, DVC) – “This is not a usual ML class; it covers the productionalization part of ML projects – the most important part from a business point of view.”
• Jeremy Jordan (Senior ML Engineer, Duo Security) – “This will be a great journey for those interested in deploying machine learning models that lead to a positive impact on the product.”

The “wall of love” on the course’s website indicates that the course is well-received for its practical approach and relevance to industry practices:

4. Full Stack Deep Learning 

About: The Full Stack Deep Learning Course is a practical PyTorch-based MLOps course that covers the entire ML lifecycle. While the course material is free for self-study, there is also the option to pay to enrol in a cohort and earn a certificate. The last update was in October 2022, as of the time of this writing. It describes the tools and frameworks you can use for the different aspects of MLOps, including their pros and cons. 

Ideal for: Python developers looking to dive into deep learning and MLOps.

Curriculum scope: This course contains the following series of lectures:

• Lecture 1: Course Vision and When to Use ML
• Lecture 2: Development Infrastructure and Tooling
• Lecture 3: Troubleshooting and Testing
• Lecture 4: Data Management
• Lecture 5: Deployment
• Lecture 6: Continual Learning
• Lecture 7: Foundation Models
• Lecture 8: ML Teams and Project Management
• Lecture 9: Ethics

Key learning objectives: 

• Understand when to use ML.
• Learn how to set up development infrastructure and tooling.
• Learn what and how to use experiment management tools, troubleshoot, and test your models.
• Learn data management and annotation concepts and tools.
• Deploy machine learning models to production and build workflows for continuous learning and model monitoring.
• Introduction to foundation models.
• Learn how to manage ML projects, ethical concerns, and how to work with ML teams.

Recommendation: This resource focuses on model development and deployment with PyTorch, arguably today’s most popular deep learning framework. This course is the top choice if you’re working with PyTorch or are most comfortable learning through hands-on coding tutorials.

Course review: The course is free and recommended for those who have already trained models and want to develop their skills further. It focuses on best practices for building AI-powered products from scratch using deep neural networks. You can visit the Full Stack Deep Learning Course website for more information and possible student testimonials.

Here’s a comparison of the four MLOps courses

5. Vendor-specific training courses by major cloud platforms

The big three cloud providers – Amazon, Microsoft, and Google – are machine-learning powerhouses. They all offer certificates for machine learning and MLOps practitioners tailored to their respective ML products.

If you’re working on one of these cloud platforms, consider taking an official course that prepares you for the certification exam:

• Preparing for Google Cloud Certification: Machine Learning Engineer Professional Certificate on Coursera
• Machine Learning Learning Plan on AWS Skill Builder or the MLOPs Engineering on AWS live classes
• Microsoft Azure Data Scientist Associate (DP-100) Professional Certificate on Coursera

All of these courses cover MLOps concepts, and you’ll complete numerous exercises.

MLOps books

While video tutorials offer on-the-go technique implementation, especially for visual learners, books provide a solid grasp of concepts alongside practical implementation. Here are the top MLOps books recommended for 2024.

1. Machine Learning Engineering with Python
2. Introducing MLOps by Mark Treveil and the Dataiku Team
3. Implementing MLOps in the Enterprise by Yaron Haviv and Noah Gift
4. Machine Learning Engineering in Action by Ben Wilson
5. Designing Machine Learning Systems by Chip Huyen
6. Practical MLOps by Noah Gift and Alfredo Deza
7. Reliable Machine Learning by Cathy Chen et al.

Jump right into the comparison of all MLOps books.

1. Machine Learning Engineering with Python: Manage the Lifecycle of Machine Learning Models Using MLOps with Practical Examples

About: Machine Learning Engineering with Python by Andrew P. McMahan is a practical guide to taking machine-learning projects to production. The author uses an example-based approach to help you develop hands-on experience in model deployment and management. Along the way, he introduces the reader to the latest open-source and proprietary MLOps tools. Packt published this book in August 2023.

Related article

Your First MLOps System: What Does Good Look Like? With Andy McMahon

Read more

Ideal for: ML engineers, data scientists, and software developers with basic knowledge of machine learning and intermediate experience with Python.

Highlights: Machine Learning Engineering with Python provides a code-centric introduction to ML engineering and an overview of the model development process. Readers learn about current deployment patterns and tools, such as LangChain. The book covers many tools like AWS, Apache Spark, Kubernetes, Ray, Apache Airflow, ZenML, Kubeflow, and GitHub Copilot through examples and projects.

Recommendation: This book introduces a wide range of tools and covers MLOps for generative AI, especially Large Language Models (LLMs).

Book reviews: Here’s a review from X’s reader:

Check reviews on Goodreads and Amazon.

2. Introducing MLOps: How to Scale Machine Learning in the Enterprise by Mark Treveil and the Dataiku Team

About: Introducing MLOps by Mark Treveil and the Dataiku Team offers guidance on ML model operationalization and team collaboration. The book covers MLOps basics, the machine learning lifecycle, and many practical examples. Despite being published in November 2020, this timeless O’Reilly book is still relevant.

Ideal for: Intermediate to advanced-level developers aiming to implement MLOps best practices in their projects.

Highlights: Introducing MLOps discusses the “What”’ and “Why” of MLOps, the key players involved in ML projects, and their roles. It provides insights into the five steps of the model life cycle – build, pre-production, deployment, monitoring, and governance – highlighting best practices for each step.

Recommendation: This book is great because it gives the learner an understanding of the intuition behind different aspects of MLOps. It comes with many real-world examples that developers can work through to get a deeper experience and add to their portfolio.

If you’re a beginner, to get the most out of this book, study it chronologically to get a well-grounded knowledge of MLOps. But if you already have experience in MLOps, you can directly jump to the chapters relevant to your current project.

Book reviews:

Here’s a review from a Goodreads reader, Charluff:

As the context gets more complex with specific cloud technologies for each part of the ML process, while business demands a lower time-to-market, decoupling, standardization and teamwork become crucial to building robust ML solutions.

As ML gets consumed via API endpoints (following microservice architectures), it is important to consider your models as software. Therefore, DevOps for ML.

This book provides a good guide to implementing a structure for your ML deliveries and processes. I’m using it to implement the ML framework at one of my customers.

The bigger the team and projects, the more you’ll benefit from this approach.

Check what people say on Goodreads and Amazon.

3. Implementing MLOps in the Enterprise: A Production-First Approach by Yaron Haviv and Noah Gift

About: Implementing MLOps in the Enterprise by Yaron Haviv and Noah Gift is a practical guide on using pre-trained models like those provided by HuggingFace and OpenAI in an enterprise setting. Released in October 2023 by O’Reilly, it was previously available through the publisher’s “Early Release” program, where it quickly found an enthusiastic audience.

Ideal for: Intermediate to advanced-level data scientists and ML engineers.

Highlights: Implementing MLOps in the Enterprise emphasizes production-oriented topics like operational pipelines, scaling, hybrid deployments, real-time predictions, and adapting to future MLOps trends. It includes a chapter that takes learners on a step-by-step practical walkthrough of building a production-grade MLOps project from A to Z.

Recommendation: This book will teach you how to design a continuous operational pipeline and automate most of the MLOps workflow. Each chapter comes with exercises.

4. Machine Learning Engineering in Action by Ben Wilson

About: Machine Learning Engineering in Action is authored by Ben Wilson, one of the minds behind Databricks and MLflow. In the book, Ben shares field-tested tips and tricks and design patterns for building ML projects that are deployable, maintainable, and secure from concept to production. In March 2022, Manning published the book.

Ideal for: This book is an excellent resource for a relatively large audience in the ML community. However, you should have some software engineering and machine learning background to follow the various practical examples, many of which come with extensive code samples.

Highlights: You will learn how to choose the right tools and techniques for your project and encounter reproducible methodologies for building stable data pipelines, efficient application workflows, and maintainable models. In addition, Machine Learning Engineering in Action covers resource usage and cost estimation, agile methodologies for fast prototyping, and software engineering best practices.

Recommendation: With close to 600 pages packed with practical wisdom, this book is probably the most extensive reference on MLOps available. It covers a wide range of topics, from team organization to code complexity to cloud operations. It does so on the strength of the author’s extensive experience in large-scale ML projects.

Book review: Here’s an in-depth review of the book:

Check out more reviews on Goodreads and Amazon.

5. Designing Machine Learning Systems: An Iterative Process for Production-Ready Applications by Chip Huyen

About: Designing Machine Learning Systems was written by Chip Huyen, the co-founder of Claypot AI, who previously worked at Nvidia, Snorkel, and Netflix. The book, published by O’Reilly In May 2022, provides a comprehensive guide to developing and deploying reliable, scalable, and adaptable models. Thanks to Huyen’s extensive experience as an educator, the book is accessible for those new to the field.

Ideal for: Beginner ML engineers and data scientists.

Highlights: Designing Machine Learning Systems covers ML system design fundamentals and the machine-learning lifecycle, from creating and processing training data over feature engineering to model serving and monitoring. Huyen also discusses operational challenges like detecting and addressing data drifts in production. Further, readers will learn how to take user experience and responsible ML into consideration.

Recommendation: This book shares an iterative framework explained through use-case examples, patterns, and system design. It is easy to read for machine-learning beginners because it starts with the basic principles of machine learning. Still, it offers much for intermediate and advanced data scientists and ML engineers. I also recommend this book if you want to get an overview of the data engineering aspect of MLOps.

Book review: Here’s on of the book reviews:

Check out more reviews on Goodreads and Amazon.

6. Practical MLOps: Operationalizing Machine Learning Models by Noah Gift and Alfredo Deza

About: Practical MLOps by Noah Gift and Alfredo Deza was released by O’Reilly in September 2021 but is still relevant today. As the subtitle suggests, it’s focused on operationalizing machine learning models on various software and hardware platforms.

Ideal for: ML engineers and those transitioning into MLOps from data science or Python backgrounds.

Highlights: Practical MLOps covers MLOps tools, AutoML, and monitoring methods. It discusses MLOps implementations on major cloud platforms (AWS, Azure, and GCP) and running models on mobile phones and specialized hardware. The book stands out in that it covers machine learning interoperability using the Open Neural Network Exchange (ONNX) format. Further, Gift and Deza encourage critical thinking by posing discussion questions that help readers develop the necessary intuition to create good MLOps solutions.

Recommendation: This book approaches MLOps from the perspective of DevOps best practices. When learning with this book, I recommend working through the code examples and exercises to practice.

Book review: Here’s a reader’s review of this book:

Check out more reviews on Goodreads and Amazon.

7. Reliable Machine Learning: Applying SRE Principles to ML in Production by Cathy Chen et al.

About: Reliable Machine Learning by Cathy Chen and colleagues is a practical book focused on reliable and efficient ML processes with a Site Reliability Engineering (SRE) approach. The book, written by seasoned SRE specialists from some of the biggest tech companies, was published by O’Reilly Media in September 2022.

Ideal for: Intermediate to advanced-level ML engineers and site reliability engineers looking to transition into MLOps.

Highlights: Topics include ML fundamentals, workflow frameworks, productionizing, troubleshooting, and effective ML team management. Reliable Machine Learning devotes an entire chapter to questions around fairness, privacy, and ML ethics.

Recommendation: This book takes a unique SRE perspective on MLOps and draws from the author’s extensive experience to derive best practices when implementing MLOps. It has several exciting case studies that will give you hands-on experience.

Book review: Here’s a reader’s review of this book:

Check out more reviews on Goodreads and Amazon.

Here’s a comparison of the seven MLOps books

Here’s a tabular comparison of the seven books based on their table of contents:

This comparison highlights the main focus areas and unique aspects of each book, providing a quick reference to understand their content and approach towards MLOps and machine learning engineering.

MLOps YouTube channels

Many data scientists and programmers will have learned the basics of their craft through watching tutorials on YouTube. In addition to introductory material, YouTube hosts a lot of advanced and expert content on MLOps.

Here are four channels I recommend you subscribe to:

1. Stanford MLSys Seminars
2. MLOps Community YouTube channel
3. DataTalksClub YouTube channel
4. AI Engineer

1. Stanford MLSys Seminars

The Stanford MLSys Seminars series, launched in Fall 2020, live-streams seminar sessions on machine learning and ML systems each Thursday. It highlights noteworthy work in ML systems to stimulate research and makes it accessible to a broader audience. If you watch through the live stream, you can ask questions in the chat. As of January 2024, the back catalog includes over 80 sessions.

2. MLOps Community YouTube channel

The MLOps Community YouTube channel is associated with the MLOps Community Slack. It shares the latest MLOps trends, best practices, challenges, and innovations through talks and interviews with leading engineers. The channel also includes all episodes of the MLOps Community podcast and hosts weekly live streams. To get a feel for the content, give their recent video on LLMs in production a try.

3. DataTalks.Club YouTube channel

The DataTalks.Club YouTube channel shares MLOps knowledge through Zoomcamps, hands-on workshops showing how to use popular MLOps tools, and open-source project spotlights. They also regularly host the DataTalksClub’s office hours and post interviews with MLOps practitioners. Check out the intro to their 2022 MLOps Zoomcamp to get an idea of what the channel is all about.

4. AI Engineer

The AI Engineer channel on YouTube publishes talks, workshops, and, more recently, recordings from the inaugural AI Engineers’ Summit held in October 2023. Check out this live stream recording to get an impression of the range of topics covered on the channel.

MLOps podcasts

If you’re on the move a lot or have a long commute, podcasts can be a great way to use that time to learn something new. There are many machine learning and MLOps podcasts out there. Most conduct interviews with MLOps professionals, which is a great way to peek behind the curtain of industry leaders’ machine learning teams.

 I recommend the following five podcasts:

1. ML Platform Podcast
2. The MLOps Community podcast
3. TWIML
4. Practical AI
5. The Machine Learning Podcast

1. ML Platform Podcast

The ML Platform Podcast is hosted by neptune.ai. It features conversations about building ML platforms and deploying them at scale. Each episode touches on various aspects of MLOps, and the guests offer valuable insights from their day-to-day work. In its first season, the podcast was called MLOps LIVE and had a slightly different format, hosting Q&A sessions where listeners could ask experts their MLOps questions.

You can find the podcast on Spotify and Appple Podcasts or watch it on YouTube. Neptune’s blog contains transcripts of most episodes. 

I recommend you start with the recent episode on Learnings From Building the ML Platform at Mailchimp, featuring Mikkio Bazeley, which covers technical topics as well as questions about career development and team structures.

2. The MLOps Community podcast

The MLOps Community podcast is hosted by Demetrios Brinkmann, the community’s founder. It features interviews with MLOps practitioners and vendors and discussions around challenges and trends in the field. You can find the podcast on Spotify and YouTube.

3. TWIML

Formerly known as “This Week in AI and ML,” TWIML, hosted by Sam Charrington, has accumulated over seven million downloads across more than 650 episodes. It features interviews with top minds across academic research, business, and consumer applications.

You can check all the episodes on the TWIML website. If I have to pick one, I recommend the episode on Building Blocks of Machine Learning at LEGO.

4. Practical AI

Practical AI is part of the Changelog podcast family and was launched in 2018. As the name suggests, hosts Daniel Whitenack and Chris Benson explore ML and MLOps from an applied and industry perspective. This podcast’s close to 250 episodes cover current trends, provide a high-level overview, or dive deep into a topic through expert interviews. You can also request that they discuss a topic or invite a guest that you’re interested in. Super cool, right?

Check out their episode on exploring and deploying generative models to get a taste.

5. The Machine Learning Podcast

The Machine Learning Podcast is hosted by Tobias Macey, whom many data engineers and Python developers will know from his Data Engineering Podcast and the Python.__init__ podcast. Each interview features an episode with an MLOps or ML practitioner discussing a specific project or use case.

For a first listen, I recommend the episode How Shopify Built A Machine Learning Platform That Encourages Experimentation, covering the Merlin ML platform.

MLOps communities

While MLOps skills are in demand, breaking into the field is not easy. The resources I’ve compiled in this article contain a lot of knowledge, but nothing beats connecting with fellow learners and practitioners.

Luckily, there are many online and offline communities where you can find like-minded MLOps enthusiasts, tap experts for help, and share your projects and achievements. Many online communities also offer job boards and provide ways to get in touch with recruiters and future colleagues directly.

Here are four online communities that I participate in:

1. MLOps Community slack
2. MLOps discord
3. DataTalks.Club
4. HuggingFace discord

I also encourage you to research if there are in-person meetups in your city.

1. MLOps Community slack

With over 20k members, the MLOps Community slack is among the biggest MLOps communities out there. There are plenty of channels dedicated to specific topics where members discuss questions, share knowledge, and connect on all things MLOps. The community regularly hosts virtual and in-person meetups all around the world. You can find recordings of many sessions on their YouTube channel that I already recommended above. 

This community is free to join and always open to new members. I recommend this community for intermediate to advanced-level MLOps practitioners looking for a place to network, initiate collaborations, search for jobs, get career advice, and participate in discussions on industry topics.

2. MLOps discord

The MLOps discord is run by Chip Hyuen, a top voice in the MLOps space who we already mentioned for her Stanford seminar and O’Reilly book. The Discord had over 20k members at the time of writing this article. If you are more of a Discord person, this community offers all the benefits you can get from the MLOps Community Slack mentioned above, like conversations with other experienced developers, career advice, and job postings. Members also post invitations to online and in-person technical events, paper reviews, workshops, and fun hangouts. 

This community is free to join and always open to new members. I recommend it for intermediate to advanced-level MLOps and data science practitioners.

3. DataTalks.Club

DataTalks.Club is a global online community of data enthusiasts discussing data engineering, machine learning, and MLOps. In the community, you can learn from free courses, participate in Zoomcamps and weekly events, and access articles and other free resources. It’s also an excellent place to get career advice. Like other big communities, DataTalks.Club shares content on a YouTube channel and blog.

This open community is free to participate in. I recommend it for all data professionals, from beginner to advanced level. Joining their Slack and signing up for the community newsletter are good first steps.

4. HuggingFace discord

The HuggingFace discord is a great place if you’re working with HuggingFace’s models and libraries and are interested in LLMs and generative AI. You can ask for help, learn along with others, collaborate on open-source projects in the HuggingFace ecosystem, discuss tools and technology, and find job advertisements.

I recommend this open community on Discord for everyone working with transformer models or interested in NLP and generative AI more broadly.

MLOps meetups and conferences

Meetups and conferences are great places to learn about the latest advancements in ML tools and techniques from experienced professionals. You can learn how MLOps techniques vary in small-scale (reasonable scale) and large-scale companies at these events. You can also have one-on-one conversations with other developers, sharing experiences and bounce ideas off each other.

Here are four conferences that you can attend virtually:

1. MLOps World: Machine Learning in Production
2. apply() by Tecton
3. MLCon
4. LLMs in Production

You should also head to your favorite search engine to see if there are local conferences in your region on the world.

1. MLOps World: Machine Learning in Production

MLOps World, initiated by the Toronto Machine Learning Society, is an international conference focused on deploying ML models in production. In 2023, the three-day conference offered hands-on workshops and talks by speakers on various topics, from business strategy to technical advancements. Virtual participation is possible. You can see many past talks on their YouTube channel.

2. apply() by Tecton

Tecton’s apply() events focus on data engineering for applied ML. Their flagship event, apply(conf), is a practitioner conference addressing ML engineering challenges ranging from tooling, labeling, and feature serving to scalability. Their events comprise workshops and in-person social events to maximize learning and networking opportunities. You can join the Tecton/Feat Slack community community to keep track of their upcoming meetups and conferences. The conference’s session and video archive is free to access.

3. MLCon

The MLCon conference series hosts events on different continents where machine learning experts and innovators share their ideas and experiences. The conferences are usually organized as hybrid events. You can check out recordings from past conferences on their YouTube channel. 

4. LLMs in Production

LLMs in Production is a series of virtual events organized by the MLOps Community. As the name suggests, it focuses on the technical aspects of deploying and managing Large Language Models. You can find recordings of the talks in dedicated playlists on the community’s YouTube channel.

Other MLOps resources

In this section, I’ve compiled a few more resources that didn’t fit the categories above.

1. ML-Ops.org and Awesome MLOps

ML-ops.org is a comprehensive guide to MLOps, offering insights into principles, components, tools, and frameworks, aiding beginners in MLOps. The extensive list of references on the website is a great place when you’re looking for in-depth material.

Awesome MLOps is a GitHub repo associated with the website that compiles an extensive list of resources around MLOps, including papers, talks, systems, ethics, and more.

Larysa Visengeriyeva, a seasoned MLOps expert, oversees both these resources. She also founded the Women in Data and AI summer festival.

2. Google Cloud’s MLOps resources

The MLOps content in Google Cloud’s documentation provides a comprehensive introduction to continuous integration (CI), continuous delivery (CD), and continuous training (CT) techniques for ML systems, which are well worth checking out even if you’re not using Google Cloud products.

3. Microsoft Azure MLOps GitHub repository

The Microsoft Azure MLOps GitHub repository contains many examples for setting up MLOps. The topics covered span model lifecycle management, training best practices, and real-world examples. Naturally, it focuses on Azure services, but many of the design patterns and project structures also translate to other tools and platforms.

Wrap up

MLOps is a big and exciting field. I hope this article has given you some ideas about where you should go next in your journey of exploring how to put ML models in production.

While you’re still here: Don’t forget to follow this blog. We regularly publish articles on MLOps, all of which you can find in the dedicated category.