The Syllabus

• Course 1: Introduction to Machine Learning
• Course 2: Developing Your First ML Workflow
• Course 3: Deep Learning Topics within Computer Vision and NLP
• Course 4: Operationalizing Machine Learning Projects on SageMaker
• Capstone Project

Introduction

In this nanodegree, Udacity attempts to provide comprehensive training in both machine learning and the AWS Cloud ecosystem. This marks my fifth nanodegree on Udacity, and I feel compelled to share an honest review, particularly because Udacity often filters out critical voices from its website.

Instructors

• Matt Maybeno
• Joseph Nicolls
• Charles Landau
• Soham Chatterjee
• Bradford Tuckfield

While these instructors use AWS and machine learning in their research and work, they are not specialized cloud experts in ML solutions.

Projects

Project 1: Predict Bike Sharing Demand with AutoGluon

This project aims to familiarize students with the open-source Auto ML library, AutoGluon.
(Good project for learning AutoGluon.)

Project 2: Build an ML Workflow for Scones Unlimited on Amazon

The goal is to build and deploy an image classification model using AWS SageMaker.
(Good project for getting started with SageMaker.)

Project 3: Image Classification Using AWS SageMaker

In this project, you use SageMaker alongside deep learning models. Unfortunately, the project is poorly designed: it cannot be completed without an inference script, which is neither mentioned in the course material nor included in the starter scripts. You’ll only discover this requirement later, and without it, the endpoint can't process images for inference.

Project 4: Operationalizing an AWS ML Project

This project is similar to the previous one but demands more details and screenshots, seemingly for review purposes.

Capstone Project

You can choose either a premade project or a custom one. The capstone consists of two parts:

• Proposal
• Implementation

Pros & Cons

Pros

• None identified.

Cons

• The nanodegree attempts to cover machine learning and the AWS ecosystem but fails at both. Content barely scratches the surface and lacks depth in every area.
• AWS documentation, tutorials, SkillBuilder, and AWS’s paid classroom lessons are all far more valuable than this nanodegree.
• The last three projects are repetitive, differing only in minor requirements.
• AWS accounts provided for SageMaker have a $25 usage limit. When this is reached, the cloud gateway becomes unresponsive. If you exhaust all credits and accounts, you must continue using your own resources.
• Each course provides a new AWS account. Be sure to keep local backups, as the platform can become inaccessible without warning.
• No career services are available for students on enterprise grants.
• Student forums lack active mentors. My own queries went unanswered, and mentor responses are rare in general.

Review Overview

Is it worth the money?
Big No.

Is it recommended for those seeking AWS Machine Learning Speciality certification?
No.

Final Thoughts

Student support has reached an all-time low. The nanodegree does not provide in-depth knowledge of SageMaker or practical machine learning skills. I do not recommend this nanodegree at all. My advice: begin by learning machine learning with on-premises resources. Once confident, use AWS documentation, tutorials, and example notebooks to learn SageMaker, and then practice directly in the cloud with the AWS Free Tier or credits provided at AWS events.