Top 10 Recommended MLOps World 2021 Sessions

MLOps World begins (soon!) on June 14, and it’s full of interesting sessions covering methods, principles and best practices for bringing ML models to production. The talks, demos and workshops will discuss feature engineering, feature stores, version management, CI/CD architecture, optimization of pipeline schedule, ML strategies, and more.

The depth and breadth of topics might make choosing which sessions you want to attend a difficult task. To help, we’ve put together our pick of the top 10 MLOps World sessions. We chose them based on the various real-life use cases they cover, the interesting stories they tell and their comprehensive approach to ML pipelines. We highly recommend checking them out. Some might also require reserving your spot in advance, so make sure to check.

We will also be there, more details below.

Here are our top 10 recommended sessions for MLOps World 2021 (in chronological order):

1. From Concept to Production: Template for the Entire ML Journey

June 14, 3:00 pm - 7:25 pm ET

Speakers: Chanchal Chatterjee, Timothy Ma and Elvin Zhu

A two-part, hands-on workshop teaching how to build and deploy ML components from concept to production-ready. The workshop is based on an open source Python template created by the presenters and will build components like data prep, model hyper train, model train, model deploy and online/batch prediction. Deployment will take place in a Kubeflow pipeline. We chose this session because it shows the entire ML journey, because it provides a practical hands-on explanation and because it is based on open source, which everyone can use.

2. Essential Workshop to Exploratory Data Analysis and Feature Engineering

June 15, 9:00 am - 1:30 pm ET

Speakers: Vladimir Rybakov and Aleksandr Mester

A problem-solving workshop focused on tackling challenges in data pre-processing and feature engineering, like noise, missing values, excessive information and more. The workshop will cover how to preprocess data, analyze it, eliminate the least useful features, create new ones and train ML models. We chose this session because it covers important parts of the ML pipeline that don’t always get a lot of attention.

3. From 12 Months to 30 Days to AI Deployment: An MLOps Journey

June 16, 11:40 am - 12:10 pm ET

Speakers: Yaron Haviv, David Aronchik and Greg Hayes

Microsoft, Iguazio and Ecolab will share the real story of how Ecolab cut down AI service deployment from 30 to 90 days by building a cloud data science architecture. We chose this session because it explains the technology and how it is implemented in organizational processes to provide real business value.

4. Systematic Approaches and Creativity: Building DoorDash's ML Platform During the Pandemic

June 16, 1:05 pm - 1:35 pm ET

Speakers: Hien Luu and Dawn Lu

The story of DoorDash and how they built an ML platform during the pandemic, while maintaining a systematic approach with clear principles but still keeping an open mind for creative ideas. We chose this session because it’s the real story of how a company built an ML platform, and under very extreme conditions at that!

5. Machine Learning on Dynamic Graphs

June 16, 1:40 pm - 2:25 pm ET

Speaker: Emanuele Rossi

An interactive session discussing Temporal Graph Networks, which is a GNN approach for machine learning of dynamic graphs, while GNN usually covers static graphs. We chose this session because it solves a pressing problem and because the solution was developed from the bottom-up at Twitter by the researchers who encountered it.

6. The AI Captain; A study of ML at the Edge

June 16, 1:40 pm - 2:25 pm ET

Speaker: Rob High

This recently added talk will be presented by Rob High, Vice President and CTO at IBM Networking and Edge Computing. Rob will take us through the Mayflower autonomous ship project, while exemplifying how AI is being brought to the edge with Edge computing. We chose this session because the Mayflower project is intriguing in its idea, its implementation and above all—its future potential.

7. Taming the Long Tail of Industrial ML Applications

June 17, 1:40 pm - 2:25 pm ET

Speaker: Savin Goyal

A session by Netflix, explaining how their data scientists improve the development and deployment experience for ML workloads in different data science aspects, from optimizing content delivery and informing buying decisions to fighting fraud. The talk will focus on their ML platform that offers abstractions for managing the model’s lifecycle from end-to-end. We chose this talk because it explains the technical-scientific aspects of ML, as well as its implementation in an organization and how to develop an ML-driven culture.

8. We Protect, They Attack: Adversarial MLOps with Elastic Security

June 17, 3:55 pm - 4:25 pm ET

Speaker: Jessica David

An interactive talk explaining how to train and deploy models on a monthly basis to enable detection and mitigation of security threats. This talk will exemplify getting from model to endpoint with various data pipelines and operational workflows. We chose this session because it provides a solution to the ever-growing and ever-changing challenges of cybersecurity threats, by implementing ML in a dynamic environment.

9. Feature Stores: Your MLOps Competitive Advantage

June 10, 12 pm - 1 pm ET

Speaker: Adi Hirschtein

A demo talk by our VP Product, Adi Hirshtein, who will explain the value of feature stores for simplifying and accelerating the path to AI in production and improving model accuracy. Adi will show how feature stores can accelerate the development and deployment of AI applications by enabling teams to build, share and manage features across the organization and across different use cases. . We chose this talk because we believe in the value of shared feature stores for the data science community .

10. MLOps Orchestration: Your Highway to Accelerating Deployment of AI

June 14, 12:00 pm - 2:00 pm ET

Speaker: Yaron Haviv

A workshop by Iguazio’s CTO, Yaron Haviv, exploring the concept of MLOps Orchestration and how it can be used for simplifying getting data science to production in any environment. Yaron will demonstrate this with the open source tool MLRun. The session will show how to map out the business problem, identify the right tools and run AI models in production. We chose this session because we know how challenging it is to get data science to production to provide real business value, and because it shows interesting use cases like fraud prediction, real-time recommendation engines and predictive maintenance.

Meet Iguazio at MLOps World

We have a number of interesting sessions at MLOps World that we hope you’ll attend and we’d love to continue hearing from you after these sessions. To share your feedback and thoughts and brainstorm ideas with us - join us on the MLOps live community on Slack for more.

See you at MLOps World!