Learning AI/ML on AWS in 2024

Learning AI/ML on AWS in 2024

Tips and Strategies for Beginners

Published Jan 17, 2024
For most of my career I’ve been interested in Artificial Intelligence (AI) and Machine Learning (ML) as a passive observer, but the idea of “learning” AI and ML was intimidating. I naively thought that AI was only accessible to PhDs and Math wizards.
But something changed recently! The BOOMING popularity of Generative AI has led to an explosion in learning resources. There's now an abundance of information, content and technologies that make learning AI with no previous experience attainable for anyone with device and a connection to the internet.
So in this post, I’m going to walk through the approach I used to teach myself AI/ML on AWS. I’ll also share specific resources that I’ve found beneficial in the AI/ML learning journey. Let’s get into it!
DISCLAIMER - I’m using “AI/ML” as a generic term to encompass ALL AI and ML concepts. This includes Generative AI, and traditional ML concepts. They are closely related technologies build on the same computing techniques, and compliment each other well in the field.

The one prerequisite. Entry-level Python skills.

I personally don’t believe it’s realistic to learn AI/ML until you first understand to code. You DO NOT need to be a coding expert, but you must first become familiar with some basic coding concepts, like Variables, Functions, Loops, etc.
If you don’t already have basic coding skills, you should spend a few weeks familiarizing yourself with a programming language.
If you’re starting from scratch, I suggest focusing on Python. Python has a good learning curve for beginners, and most AI code is written in python today. Because of that you will need to read many “Code Examples” written in python along your AI/ML journey.
Don’t just take it from me. You can take it from the popular LLM, Claude 2 from AWS partner Anthropic
“Machine learning can seem complex at first
but is really quite intuitive
once you play around with some code examples.” - Claude2
When I first taught myself python, I relied heavily on AWS lambda workshops and YouTube tutorials. Here are some handy resources to help you get started with python!
Remember, you don't need to be an expert. You just need enough python knowledge to understand how to interact with APIe, and how to understand ML code samples.
  • AWS Python Tutorials** - Check out the “Build” section for hands-on python learning exercises
  • AWS Bedrock & Langchain Workshop** - This is a great intermediate course to explore very basic AI development conepts in python.
  • Python Lab: Lab - Python Challenge** - provides you practical exercises designed to reinforce Python understanding and proficiency in a professional context.
  • AWS Lambda Workshop - Because of lambda’s simplicity and integrated IDE, it is a perfect service for beginner developers to learn on. This is the service that I wrote my very first python projects on because of its low cost and deep integration with AWS python SDKs

Tools in your Learning Kit
Outside of formal training and online courses, I’ve found these two tools invaluable in streamlining the learning process:

Personalized AI Tutor - Your AI Doesn't Mind Repeating Itself

We are lucky that LLMs like Anthropic’s Claude have been trained on mountains of Python and Machine Learning data. This means that LLMs trained on Programming and ML data-sets are leading AI experts!!!
So my learning has been super-charged by using an AI assistant. I don’t think I could have jumped in head first like I did without the help of my AI to fill in the blanks along the way.
My primary AI assistant is actually a partyrock.aws app backed by Claude Instant, which is a high speed and resource efficient LLM. Partyrock is a fun way to build AI applications in seconds, and is really useful for experimenting with basic prompt engineering. I have a prompt that tells Partyrock something like “You are a helpful assistant who’s purpose is to answer any questions about AI/ML and python. Please always include code samples where applicable. You should answer questions in plain, clear language as if you are explaining these concepts to a beginner. Avoid excessive jargon . If you don’t know the answer with high confidence, say you do not know."
In addition to Partyrock, I often use Amazon Q when debugging code errors in my IDE, or if I have AWS specific service questions.
Also CodeWhisperer is my default real-time python coding companion. It works great in VS Code along with the Q extension!

Online Videos on Math and ML Concepts

When I get ‘stuck’ on a concept, I found myself gravitating towards online videos to explain a concept. Because statistics and data are such visual topics, I found that watching short video clips in math and ML concepts helps me get past any roadblocks that my AI tutor could not.
For example, I had to freshen up on basic statistics concepts like distributions, normalization, algorithms, etc. I found visuals via popular video platforms and community blogs to be extremely helpful for the comprehension of these topics.

Learn & Apply The Theory

In addition to hands-on learning, you should make time to understand the theory behind key AI concepts. This will help you learn which AI tools are best for different scenarios. I found learning the theory to be an exciting part of the journey. Get ready for a bunch of ‘lightbulb’ moments as you immerse yourself in the AI and ML world!!
The #1 most useful tool for me in learning the theory has been preparing for the AWS Certified Machine Learning - Specialty exam. The AWS exam guide served as a compass to help me understand which ML concepts to focus on learning. This certification prepares you for common scenarios that you will run into as an ML practitioner, and will help you in building instincts for selecting “the right tool for the job” in your work.
In my opinion, if you’re going to learn something hard, you may as well get a credential out of it! So I encourage anyone taking the plunge into the ML field to consider taking the AWS Certified Machine Learning - Specialty exam within ~12 months of diving into the tech, but not until after you get experience with python and AWS development
The resources I found most helpful in developing theoretical knowledge of ML and AI were:

Applying Your Knowledge on AWS

Amazon Bedrock is an AI developer’s best friend. The service grants access to LLMs and image models from leading Foundational Model providers like Amazon, Anthropic, Meta, AI21 Labs, and more! This services helps you move extremely fast, and integrates with popular AI orchestration frameworks like LangChain.
Because bedrock is serverless, you can develop complex AI applications for low development cost. For example, I’ve spent a total of $5 last month building ~3 AI projects on the Bedrock platform so far.

SageMaker is an AWS ML platform for development, training, testing and deployment of models. ML students should familiarize with SageMaker offerings, because it will give key insight into what is needed to run ML operations at scale within a company.
SageMaker comes with a 2 month trail that allows you to experiment and familiarize with the service. If you still need access to SageMaker after the free trial, you should consider a SkillBuilder subscription. Skillbuilder gives access to hands-on-lab environments for a monthly subscription of $29; this automatically caps your spend and ensures that you won’t incur excessive Sagemaker costs while you’re still learning.

AWS also provides AI managed services that can be used to support AI development. They offer everything from pre-trained models to services that help extract and process data. Some examples are Polly, Textract, Rekognition, and Forecast. Most of the services are easy to learn and have a free tier, so its worth spending a few afternoons exploring some of these in the AWS console to understand what tools are available.
There is an abundance of free and paid training for Bedrock and SageMaker spread through AWS Workshops, AWS Skill Builder, and blogs/videos online. And at this point in your journey, you should know how to do your own digging, and find the resources that work best for you!

Start building, and think BIG

The democratization of AI knowledge means that YOU might be the next person to launch an AI initiative at your company, or build a successful AI startup.
So when you have an idea, follow that idea. Don’t be afraid to try out new things. If it feels like a big problem that nobody has solved yet with AI, you might just be onto something!!
If you need help, or have questions, please engage the community! AWS re:Post is a GREAT place to connect with other expert. If you are on LinkedIn or X(Twitter) you find find me @TrevSpires - Senior Solutions Architect @ AWS FinTech

1 Comment