O'Reilly logo
live online training icon Live Online training

TensorFlow Extended: Model build, analysis, and serving

enter image description here

Build an end-to-end machine learning pipeline with TFX

Armen Donigian

Companies are looking for ways to incorporate machine learning into their business to lower costs and increase revenue, but their machine-learned models are often parts of complex systems comprising a large number of data sources and interacting components, which are commonly entangled together. This creates large surfaces on which bugs can grow and unexpected interactions can develop, potentially to the detriment of end-user experiences via the degradation of the machine-learned model. These issues can be difficult for humans to detect, especially in a continuous training setting, where new models are refreshed and pushed to production frequently.

Join expert Armen Donigian to gain hands-on practical experience designing and transforming features, experimenting, and analyzing, serving, and profiling machine learning models using the recently open-sourced TensorFlow Extended (TFX), which allows you to leverage the state-of-the-art technology that powers most of Google’s ML systems to solve your particular business or scientific problems. You'll also explore the TensorFlow Serving framework, designed to be flexible and support new algorithms and experiments. Over three hours, you'll learn how to implement a reusable component that automatically evaluates and validates models to ensure that they are “good” before serving them to users as well as a complete serving solution for machine-learned models to be deployed in production environments.

What you'll learn-and how you can apply it

By the end of this live, hands-on, online course, you’ll understand:

  • How to analyze your model results using TensorFlow Model Analysis
  • How to evaluate your model performance using TensorFlow Model Evaluation
  • How to serve your ML models using TensorFlow Serving
  • How to use TensorFlow Profiler and TensorBoard to better understand characteristics of your model

And you’ll be able to:

  • Apply the same design and implementation principles from the technology that made Google successful to your specific project
  • Develop an end-to-end machine learning pipeline for supervised learning projects using TensorFlow Extended
  • Get hands-on experience with an end-to-end example integrating various parts of TensorFlow and make it part of your workflow

This training course is for you because...

  • You're a data scientist, business analyst, or machine learning engineer who needs to leverage machine learning to solve a specific business problem.
  • You want to learn how to build an end-to-end machine learning pipeline and release it into production.


  • A working knowledge of TensorFlow or another machine learning framework, such as scikit-learn or PyTorch

Recommended preparation:

Recommended follow-up:

About your instructor

  • Armen Donigian has undergraduate and graduate degrees in Computer Science from UCLA and USC. He started his career building tracking & navigation algorithms at Orincon (later acquired by Lockheed Martin). Armen then joined the Global Differential GPS group at Jet Propulsion Laboratory (NASA), performing clock and orbit corrections using GPS/GLONASS satellites, which were also used for testing of Mars Science Laboratory Curiosity Rover.

    Bitten by the startup bug, Armen has helped several startups build data driven products and scale infrastructure as a Senior Data & Machine Learning Engineer. Armen has previously led the development of machine learning explainability methods & currently works as the Head of Personalization & Recommender Systems at Honey Science.


The timeframes are only estimates and may vary according to how the class is progressing

An overview of the problems TFX can help you solve (30 minutes)

  • Lecture: TensorFlow Estimator, Model Analysis, and Serving
  • Hands-on exercise: Knowledge check

Case study: An end-to-end example integrating various parts of the TensorFlow ecosystem together (15 minutes)

  • Lecture: An end-to-end notebook demonstrating TensorFlow Estimator, Model Analysis, and Serving
  • Hands-on exercise: Knowledge check

Break (10 minutes)

Model training with TensorFlow Estimator (40 minutes)

  • Lecture: Using a premade Estimator to build your first model and use its results to establish a baseline; building and testing your overall pipeline, including the integrity and reliability of your data with this premade Estimator; finding suitable alternative premade Estimators and running experiments to determine which produces the best results; improving your model by building your own custom Estimator
  • Hands-on exercise: Knowledge check

How to analyze your model results using TensorFlow Model Analysis (40 minutes)

  • Lecture: How to slice metrics; various visualization plots available; how to specify custom metrics; multiple model analysis; multiple data analysis
  • Hands-on exercise: Knowledge check

Break (5 minutes)

How to serve your ML models using TensorFlow Serving (40 minutes)

  • Lecture: Training and exporting a TensorFlow model; managing model versioning with TensorFlow Serving ServerCore; configuring batching using SessionBundleSourceAdapterConfig; serving a request with TensorFlow Serving ServerCore; running and testing the service
  • Hands-on exercise: Knowledge check