Abstract: This will be a live workshop. working in real-time with tools, deploying models into production In our ODSC talk, we present an MLOps manifesto, proposing an architecture and a set of open-source tools to make ML reproducible, accountable, collaborative and continuous. In this workshop, we will show we will implement this manifesto by giving a detailed and candid demonstration of tools. The target audience is data scientists, machine learning engineers, and technical managers who want to get their models into production in an enterprise environment.
In this workshop, we will show and implement our MLOps manifesto by giving a detailed and candid demonstration of tools. we will cover
Reproducibility: ""Version everything"": datasets (they can be large), models, parameters, notebooks, and environments
Accountability: Runs, provenance graph of data & models
Collaboration: Share projects, merge and diff Jupyter notebooks, asynchronous collaboration
Continuous delivery: Integration of continuous integration and delivery (CI/CD) tools
Deployment: Deployment of models via Docker to Kubernetes
Monitoring: Statistical model monitoring via the Prometheus time-series database, and visualization with Grafana
Run anywhere: On the cloud, on-premise, or hybrid.
Bio: Luke Marsden, CEO and founder of Dotscience, is an industry veteran and an expert in DevOps and container storage. He founded Dotscience on the belief that operationalizing Machine Learning should be just as easy, fast and safe as modern software engineering became when DevOps revolutionized the industry.
Prior to becoming the founder and CEO of Dotscience in 2017, Marsden was the Head of Developer Experience at Weaveworks, where he worked to optimize and revolutionize the development process for open source users. Before his time at Weaveworks, Marsden founded the container storage startup ClusterHQ and served as the startup’s chief technology officer.
A thought leader in his field, Luke has developed experimental software, regularly contributes to the open source community and often speaks at meetups and prominent industry conferences.