Online Applied AI & DevOps in partnership with QuantumBlack
Details
Our next Meetup is going virtual! Our last three events were a resounding success with over 100 people in attendance, so we thought we should continue to grow our great tech community by partnering with QuantumBlack for our first-ever virtual event.
The main theme will be ‘Machine Learning in Production’ so expect some insightful talks from our presenters.
The great thing about taking it virtual is that you can ask questions throughout the session - in between each talk we will allow some time for the speakers to answer some of your questions.
We do hope you can join!
WEBINAR LINK HERE - https://quantumblack.zoom.us/j/94516792505
Evening itinerary:
6:30 pm - 6:35 pm - Introduction for the evening
6:35 pm - 6:50 pm - Shubham Agrawal, Machine Learning Engineer at QuantumBlack - https://www.linkedin.com/in/agrawalshubham1729/
Taking AI from Prototype to Production
Shubham will explore the differences between deploying Machine Learning for industry and academia, bringing the discussion to life through a series of QuantumBlack projects from across the pharma, renewable energy and motor racing sectors. He will also outline the lifecycle of a typical ML project, offering practical advice on developing production-ready code.
6:50 pm - 7:05 pm - Mattias Arro, Machine Learning Engineer - https://www.linkedin.com/in/mattias-arro-b63a9110/
Introduction to Kubeflow
Mattias will be discussing an overview of the rapidly growing ML productionisation solution Kubeflow. Kubeflow is a set of Kubernetes services for building data pipelines, model training and deployment systems, and exploratory ML work.
The talk will cover the main tools in the Kubeflow ecosystem, and how to deploy and maintain a Kubeflow cluster.
7:05 pm - 7:20 pm - Jan Teichmann, Senior Data Science Consultant at Trainline - https://www.linkedin.com/in/janteichmann/
Solving the hardest problem in data science: Production deployment
Making data science a success is really hard with up to 85% of projects failing, according to Gartner. While the reasons are complex, the most challenging problem in data science remains production deployment of models. What is so different about data that it needs new approaches?
This talk will explain the two types of model deployment in production and their pros and cons with solutions straight from the front lines.
I look forward to seeing you all there :)