This course includes 8 hands on workshops, and each workshop will cover:

A part of the ML production lifecycle using GCP native tools along with an example business use case and how AI maps to that use case

An AI/ML algorithm and AI/ML internals technical topic. For example, how and when to use Regression, Classification and how to map the right powerful AI algorithms for each data type.

How to build a production machine learning services (Feature Engineering, Model Training, Model Validation, Endpoints, Gateway/Cloud Functions Integration, Application Integration).

One or more GCP tools. Across the 8 sessions we will cover GCP Storage, GCP AI Platform BuiltIn Algorithms, GCP AI Platform with Custom algorithms, GCP AI Platform models, GCP Cloud functions, GCP Roles and Authentication, GCP Logging. We will also cover additional tools - Postman API testing, cURL, Production AI Best Practices, Microservice design pattern for AI deployment.

You will build 7 working production ML services.

You will build a custom project with a problem you choose

We will cover the following projects (business use cases):

Customer Churn: Detect whether your customers are about leave your business or your product. Predict which customers are at risk of churn.

Pricing Analysis: Understand what factors most affect price. Predict how price can change with features. Assess viability of price for future products

Customer approvals: Should you approve a loan for a particular customer? Predict whether a customer is likely to be delinquent on a bill?

Appointment planning: Is a customer likely to miss an appointment? Can you plan your appointment schedule more effectively?

Removing bias: Bias in your AI data can lead to poor outcomes, unhappy users or even legal problems. How to detect and remove sources of bias?

Sentiment Analysis: Are you customer’s happy with your product? What do their comments, tweets and other writings say about their feelings?

Text Classification: What can you learn about your customers from the reviews they provide using the IMDB reviews dataset

You will learn how to use cloud tools:

Concept of projects in GCP

Configure and use GCP Storage for your data

Feature engineer your data with Python code

Configure GCP AI Platform Deep dive on built in GCP AI Platform algorithms Linear Learner and XGBoost. How to hyper-parameter tune AI Platform algorithms.

Bring custom code into GCP AI Platform

Configuring and using a GCP AI Platform generated models

Connecting a GCP AI Platform models to a public URL via Cloud functions

Integrating REST microservices with applications. Using CURL/Postman for API testing.

Cloud GCP best practices. Logging and managing model versions.

You will learn machine learning algorithms, algorithm internals and technical concepts

How to select an algorithm for a use case, train, deploy and use it in production, and measure how well it is doing.

How to map each use case to machine learning - what type of machine learning can be used, for what data type. How to measure.

Powerful general purpose algorithms like XGBoost, LinearLearner, KNN, etc. and how they work internally and how to hyper-parameter tune them for best performance.

How to build algorithms for natural language processing in the context of classification

Metrics and practices for algorithmic evaluation and how to map them to use case.

AI Trust, Fairness and Bias. Managing AI related risks

Machine learning explainability

Advanced aspects of production ML - live monitoring and diagnosis, model versioning, retraining and others.

In this first session, we will introduce the stages of the pipeline for building production AI in the cloud, along with the GCP tools used for each stage. We will describe our first business use case, Customer Churn. In the code lab, the attendees will use GCP Storage to upload the Churn dataset and get started with building the AI pipeline.
What students will learn:

• Overview of the production ML lifecycle and all of its steps. How to go from data to running production ML
• Description of cloud services that can be used for each stage
• Overview of a customer churn problem and how to build a ML service for it with GCP with Binary Classification Algorithms
• Lab 1: Build a ML service for Customer Churn and test it.

This session builds upon the first. We will go over an important aspect of the ML pipeline, Feature Engineering. We will go over the transformations that are required before a dataset can be used by an GCP ML algorithm for building a model. We also examine the use case of Pricing Analysis and how a production AI based on Regression techniques can address this use case. We follow the lifecycle steps shown in the first session, but delve into more depth on how to perform appropriate feature engineering for different scenarios. We also introduce attendees to an ML algorithm called XGBoost and how it can be used for both regression and classification problems. In the code lab, attendees will perform feature engineering transforms, save the transformed data back in GCP Storage and get started with configuring a XGBoost algorithm in GCP AI Platform.
What students will learn:

• How to perform feature engineering as a part of the AI life cycle in production
• Overview of a business Pricing Problem and how to build an AI for it using Regression Algorithms
• Common approaches to Feature Engineering - One Hot Encoding and Missing Value handling
• Lab 2: Perform feature engineering on two datasets, Pricing and Churn. Build an AI in the GCP

In this session, we will explore a new algorithm called Linear Learner and configure its hyper-parameters in GCP AI Platform. We will also learn how to evaluate the performance of an AI algorithm and use this criterion to tune the hyper-parameters and pick the best model. In particular, we delve into evaluation metrics beyond accuracy and cover both ML metrics (such as Confusion Matrix, RMSE) and production service metrics (such as latency and scale) that are important for production. As a case study, we will use the Pricing and Churn datasets that were introduced in the previous sessions
What stuents will learn:

• How to build and configure hyper-parameters for both regression and classification type of problems
• Understand the hyper-parameters exposed by GCP AI Platform and use Linear Learner and XGBoost for performing training on the two types of problems: Classification and Regression
• How hyper-parameters affect solution quality and performance
• Lab 3: Train and evaluate several ML algorithms in the cloud using GCP AI Platform and GCP Logging

In this session, we will focus on deployment of the trained ML model into production as a microservice. This will involve several aspects, such as endpoint creation, IAM role creation and configuration of a Cloud Function deployment. This session will focus on the application integration of production AI and how to test and evaluate this integration. In the code lab, attendees will build an end to end working AI and perform external evaluation of the AI to access its performance.
What students will learn:

• How to deploy an AI in the cloud as a microservice
• Evaluate a production AI with both ML metrics (like Accuracy, Confusion Matrix, True/False Positive/Negative etc.) and code metrics (throughput, accuracy, etc.)
• How to test their AI service using CURL and Postman
• How to integrate their AI service into Python, Java or Javascript applications
• Lab 4: Build an end to end production pipeline in the cloud

In this session, we will spend a part of the time to finish up the end to end lifecycle using the custom algorithm described in the previous session as it typically takes more than one session to cover. After that, we focus on the critical issue of data Skew in AI. and AI Bias. We cover how skew in data impacts the performance of AI and steps that can be taken to eliminate it. We will also cover the common ways that Bias can enter a production AI implementation and how to detect and remove Bias. We cover in depth another powerful topic called Drift and its impact on real world production deployments. As a case study, we use the publicly available Appointment planning and COMPAS dataset. In the code lab, attendees will build several AIs using these datasets. We will also introduce the overlay tool Pyxeda Navigator and provide resources for attendees to use it
What students will learn:

• What is data Skew and how to overcome it.
• AI Bias and why should it be avoided?
• How can AI Bias enter? How can it be detected and removed?
• How to test for Bias and apply Feature Engineering to reduce it.
• Lab 5: build several AIs and perform analysis to detect skew and bias

In this session, we will focus on using the Natural Language service provided by GCP. We dive deep into algorithms for Text Classification and the basics of the pre-processing pipeline for text type of data. We given an overview of basic concepts like Bag of Words approach, its operation, advantages and limitations and compare it with state of art such as BERT. We will look into how GCP AI Platform allows the deployment of NLP algorithms, such as text classification. Attendees will build an end to end working AI for NLP.
What students will learn:

• How to build and evaluate production AIs that use Text Classification algorithms
• Overview of a business Sentiment Analysis problem and how to build an AI for it using Text Classification in either Binary or Multiclass form
• How to pre-process text data so that it conforms to GCP requirements
• Lab 5: build an end to end production pipeline in the cloud using custom algorithms.

In this session we show how one can train an NLP algorithm using GCP Natural Language service on custom data. We will train the algorithm and evaluate its performance. We will also deploy the trained model as a microservice. In the code lab, attendees will build a review predictor in the production system using the publicly available IMDB data, illustrating the concepts covered so far
What students will learn:

• A methodology for mapping use cases (with different data types) into AI algorithms and lifecycle steps
• Processing the public data to train using GCP
• Metrics to evaluate the custom modell
• Lab 7: Build and use a production AI for review predictions.

We will provide an overview of advanced topics that attendees can explore beyond this webinar series. These include Model Versioning, interactions between application versioning and model versioning, diagnostics of production AI in the presence of data changes, A/B testing, and others. In the code lab, students will complete their custom project and create a github and project repository to showcase their project. They can also create a project video if they choose to and put it in a hosting facility that we will provide
What students will learn:

• Best practices for Production AI (model versioning, model integrity, retraining cycles, microservice API management and versioning). How to manage production application changes and production AI upgrades in concert
• Best practices (debugging, instances, your cloud bill :-)).
• Lab 8: Finish custom project and projects presentation

Nisha&Sindhu

Sindhu Ghanta
Head of Machine Learning in Pyxeda. She was a Post-Doctoral Fellow with BIDMC and the Department of Pathology, Harvard Medical School, where she was involved in detection and classification of features from histopathological (breast cancer) images. She worked as a research scientist with Parallel Machines on monitoring the health of machine learning algorithms in production and has many publications on ML innovations.

Nisha Talagala
Founder of Pyxeda AI. Previously, Nisha co-founded ParallelM which pioneered the MLOps practice of managing machine learning in production.

Yamuna Dulanjani
Senior Machine Learning Engineer at Pyxeda AI