What’s Trending in MLOps in 2022?
A model that never makes it into production is one that is incapable of producing value for a business or organization. Unfortunately, the percentage of models that make it out of development is still low. However, the field of MLOps is focused on this very problem and has come up with a number of solutions, best practices, and tools to help organizations effectively integrate machine learning and AI models into their business practices. These MLOps trends will be helpful beyond just 2022.
To help you learn the tools and skills you need to implement MLOps in your organization, ODSC East 2022 will feature talks, workshops, and training sessions led by some of the best and brightest minds in the field. Check out some of the topics, tools, and use cases that will be discussed below. These MLOps trends go beyond the conference, and show what’s trending in 2022, who to pay attention to, what companies are excelling with MLOps, and emerging tools.
Engineering a Performant Machine Learning Pipeline: From Dask to Kubeflow
Dr. Rahul Dave | Chief Scientist | univ.ai, lxprior.com, Harvard University and Richard Kim | Founder and CEO | Markov Lab
Machine learning pipelines are essential for any organization that wants to integrate AI into its business practices. Pipelines, which automate aspects of the machine learning workflow, not only increase efficiency, but also free up your data scientists and AI practitioners to work on higher-level projects. During this ODSC East session, you’ll learn how to engineer a performant machine learning pipeline from two MLOps experts.
AI Observability: How To Fix Issues With Your ML Model
Danny D. Leybzon | MLOps Architect | WhyLabs
Because of the speed of change of data and information, as soon as a model is deployed, its performance starts to deteriorate. To optimize the performance of these models, therefore, it’s essential that organizations are able to see and analyze where the models are breaking down.
This talk will take you through the best practices of model observation and help you gain the knowledge necessary to keep your models performing at the highest level.
MLOps: Relieving Technical Debt in ML with MLflow, Delta and Databricks
Sean Owen | Principal ML Solutions Architect | Databricks and Yinxi Zhang, PhD | Senior Data Scientist | Databricks
Databricks was founded in 2013 and is the first cloud-based data lakehouse platform. Databricks enables organizations to combine data and AI in one unified platform, as well as offering tools that help solve the challenges of MLOps, such as Delta, open source MLFlow, and a Feature Store. Learn more about these tools during this upcoming ODSC East 2022 session.
Drift Detection in Structured and Unstructured Data
Keegan Hines, PhD | VP of ML, Adjunct Professor, Chair | ArthurAI, Georgetown, CAMLIS
To provide actionable insights that benefit a company or organization, the data supplied to machine learning models need to accurately reflect the current state of the world. However, due to the rapid pace of change, it’s difficult to ensure your data is an accurate representation of the real world, resulting in model degradation known as data drift. In this ODSC East session, you’ll learn about different strategies for handling data drift and making sure your models are optimized.
Building and Deploying Machine Learning Models with TensorFlow and Keras
Yong Tang, PhD | Director of Engineering | MobileIron
TensorFlow and Keras are some of the most widely used frameworks for machine learning. Its ability to scale with your project and its robustness make it a natural choice for many, especially when exploring new applications. Featuring hands-on examples, this session will help you understand why the TensorFlow and Keras framework has become so popular for these types of projects.
Data Science in the Cloud-Native Era
Yuan Tang | Founding Engineer, Co-chair | Akuity, Kubeflow
Data clouds have significantly changed the way that companies share data both internally and externally. This seamless sharing of data has the ability to increase efficiency, but it has also introduced new challenges for MLOps. To help you overcome these challenges, this session will focus on the available tools and best practices for MLOps for data clouds.
Introducing Model Validation Toolkit
Alex Eftimiades | Senior Data Scientist | FINRA and Matt Gillett | Software Development Engineer In Test | FINRA
To ensure that your model is ready for deployment, there are some essential questions to answer such as, how to measure things like data drift, false negatives, and confidence intervals. FINRA’s Model Validation Toolkit can help organizations answer these questions and more to make sure that their models are capable of providing useful and impactful insights.
Quick to Production with the Best of Both Spark and TensorFlow
Ronny Mathew | Senior Data Scientist | Rue Gilt Groupe
Although TensorFlow is one of the most popular frameworks to use for machine learning, the first iteration of the framework had difficulty handling big data. With TensorFlow 2, you can now incorporate Spark, solving many of the issues caused by large datasets. In this session you’ll learn how to utilize TensorFlow and Spark to handle the more challenging aspects of building models with large datasets.
Tower of Babel: Making Apache Spark, Apache Mahout, Kubeflow, and Kubernetes Play Nice
Trevor Grant | Director of Developer Relation | Arrikto
In this session, Trevor Grant will illustrate how you can use the strengths of Apache Spark, Apache Mahout, Kubeflow, and Kubernetes to cover for the others’ weaknesses for applications in healthcare that can improve efficiency and ease the strain on the limited resources of hospitals.
What We’ve Learned Pushing Nearly 100M Hours of GPU Compute
Daniel Kobran | COO and Co-Founder | Paperspace
Founded in 2014, Paperspace offers cloud computing services to tens of thousands of clients, including enterprises, individuals, and startups. As a result, they have had the opportunity to observe how organizations and people utilize MLOps, the reasoning behind these common behaviors, and the ways to improve them. In this session, Paperspace will share these insights and discuss best practices for MLOps.
Learn About MLOps Trends at ODSC East 2022
Check out these talks, workshops, training sessions, and much more, including sessions on machine learning, deep learning, NLP, MLOps trends, and machine learning safety and security at ODSC East this April 19th-21st. Register now to save.
Read more data science articles on OpenDataScience.com, including tutorials and guides from beginner to advanced levels! Subscribe to our weekly newsletter here and receive the latest news every Thursday. You can also get data science training on-demand wherever you are with our Ai+ Training platform. Subscribe to our fast-growing Medium Publication too, the ODSC Journal, and inquire about becoming a writer.
