From Machine Learning Model to Real-World Application: What I’m Learning with AWS
As an AI/ML student, I’m learning that building a model is only one part of the journey. This article shares my understanding of how cloud services can help store data, train models, deploy them, and turn an ML experiment into a usable application.
From Machine Learning Model to Real-World Application: What I’m Learning with AWS
As an AI/ML student, most of my early learning focused on algorithms, datasets, model training, and measuring accuracy.
But while learning AWS, I started understanding that creating a machine learning model is only one part of building a real application.
A useful ML system also needs data storage, computing resources, deployment, monitoring, and a way for users or other applications to interact with the model.
From a notebook to an application
A machine learning project often starts with a notebook on a local computer.
I can load a dataset, preprocess it, train a model, and evaluate its performance. However, the project becomes more challenging when I want other people to use that model.
This is where cloud infrastructure becomes useful.
For example, datasets and files can be stored using Amazon S3, while computing resources can be provided by services such as Amazon EC2.
Where machine learning services help
One service I am particularly interested in learning is Amazon SageMaker.
My understanding so far is that SageMaker is designed to support different stages of the machine learning workflow, including developing, training, and deploying models.
This is interesting because instead of thinking only about:
Dataset → Model → Accuracy
I am starting to think about:
Data → Training → Evaluation → Deployment → Real-world usage
That change in perspective is one of the most useful things I have learned while exploring AWS.
Why deployment matters
A model can perform very well during testing and still be difficult to use in practice.
For example, imagine building an image classification model that identifies different types of objects. The model itself is not enough. A complete application may need an interface, an API, cloud storage, computing resources, and a reliable deployment process.
This is why I think learning cloud technologies alongside AI/ML is valuable.
My next goal
My next goal is to build a small end-to-end AI/ML project on AWS.
I want to take a model from a local development environment and understand how it can be deployed and accessed as part of an application.
I am still learning the fundamentals, but this journey has helped me understand an important idea:
Building an ML model is only the beginning. Building a usable ML application is the bigger challenge.
I’m looking forward to learning more about Amazon SageMaker, generative AI, and AWS deployment workflows.
What was your first experience deploying an ML model outside your local environment?
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article