Full-Stack Machine Learning with Metaflow
Machine Learning is more than just algorithms, it's ultimately about delivering projects that unlock real value for a company or organization. This course presents a hands-on introduction to production-grade tools that bridge the gap between laptop data science and production ML workflows, all with an eye towards maximizing value and robustness of the outputs. Participants will learn how to productionize ML workflows using Metaflow, while thinking strategically about how their work fits into the broader landscape of a company's business and technology strategy.
Course taught by expert instructors
Ville Tuulos
CEO, Outerbounds
Ville Tuulos has been developing infrastructure for machine learning for over two decades. He has worked as an ML researcher in academia and as a leader at a number of companies, including Netflix where he led the ML infrastructure team that created Metaflow, a popular open-source framework for data science infrastructure. He is the co-founder and CEO of Outerbounds, a company developing modern human-centric ML. He is also the author of Effective Data Science Infrastructure, published by Manning.
Hugo Bowne-Anderson
Head of Developer Relations at Outerbounds
Hugo Bowne-Anderson is Head of Developer Relations at Outerbounds. He is also the host of the industry podcast Vanishing Gradients. Hugo is a data scientist, educator, evangelist, content marketer, and data strategy consultant, with extensive experience at Coiled, a company that makes it simple for organizations to scale their data science seamlessly, and DataCamp, the online education platform for all things data. He also has experience teaching basic to advanced data science topics at institutions such as Yale University and Cold Spring Harbor Laboratory, conferences such as SciPy, PyCon, and ODSC and with organizations such as Data Carpentry.
The course
Learn and apply skills with real-world projects.
ML Engineers and Data Scientists who want to take common machine learning models and productionize them with Metaflow.
Software engineers who want to build production systems that integrate ML.
Programming fundamentals and Python basics - variables, for loops (as covered in CoRise Python Crash Course or elsewhere)
Foundation in numpy, pandas, scikit-learn (as covered in CoRise Intermediate Python for Data Science or elsewhere)
Familiarity with terminal/shell and Jupyter Notebooks
Try these prep courses first
- Learn
- Understand the problem space: business, organizational/cultural, and technical concerns
- The dataflow paradigm and DAGs
- The basics of Metaflow
- Product Manager POV: Project Scope & Measuring Success
- Baseline model workflows for common application (RecSys, NLP, or CV)
- Learn
- Thinking from Engineer POV: what does it mean to build this application from an engineering context?
- How to build ML workflows with Metaflow
- Versioning, model reporting, and notebooks
- Iterative approach to building ML workflows
- Engineer POV: De-risking & Contingency Planning
- Hands-on experience building a reproducible ML workflow
- Learn
- Thinking from data engineering POV: what if we have 100GB of potentially messy data? Questions around ETL and interacting with data warehouses
- How to interact with real-world size datasets in a variety of common formats
- Sending particular steps in your ML workflow to the cloud using Metaflow (e.g. large training steps)
- How to handle failures with Metaflow
- Sending entire workflows to the cloud
- Issues of dependency management
- Data Engineering POV: Case Study/Scenario
- Large-scale parallel training utilizing cloud compute
- Learn
- Production ML is a spectrum involving all layers of the stack (e.g. data, compute, versioning, and so on): create MVP production deployments early on
- Understand production can mean many things: e.g. model hosting, writing results to a DB, building a report as a doc/email/slide deck.
- How to deploy a baseline model, build a challenger model, and promote it to production using Metaflow
- This will involve understanding how to measure success and how to report on it.
- Build and deploy an end-to-end ML model
Real-world projects
Work on projects that bring your learning to life.
Made to be directly applicable in your work.
Live access to experts
Sessions and Q&As with our expert instructors, along with real-world projects.
Network & community
Core reviews a study groups. Share experiences and learn alongside a global network of professionals.
Support & accountability
We have a system in place to make sure you complete the course, and to help nudge you along the way.