AWS Builder Center
From Python to Generative AI: My Journey Toward Becoming an AI Engineer

From Python to Generative AI: My Journey Toward Becoming an AI Engineer

My journey into AI started with Python and gradually evolved into machine learning, deep learning, Generative AI, and cloud technologies. In this article, I share what I’ve learned through hands-on projects, why I became interested in LLMs and RAG, how AWS fits into my AI engineering journey, and the skills I’m focusing on next.I also share practical lessons I’ve learned about moving beyond tutorials,building real projects,understanding AI systems,and developing the mindset needed to grow as an AI Engineer.

From Python to Generative AI: My Journey Toward Becoming an AI Engineer

Introduction

Artificial Intelligence has moved from being a futuristic concept to becoming a technology that developers and students can experiment with every day. As an Artificial Intelligence and Machine Learning student, I initially found the field exciting but also overwhelming. There were so many technologies to learn—Python, machine learning, deep learning, transformers, large language models, cloud computing, and AI agents.
Instead of trying to learn everything at once, I decided to approach AI engineering as a journey: build strong fundamentals, create projects, learn from mistakes, and gradually move toward modern Generative AI applications.
This article shares my journey so far, the lessons I have learned, and how I am exploring AWS and the broader AI ecosystem while working toward becoming an AI Engineer.

Starting With the Fundamentals

My AI journey started with Python.
Python is one of the most important tools in my learning path because it provides the foundation for working with data, machine learning models, APIs, and modern AI frameworks.
After learning Python, I moved toward libraries such as NumPy and Pandas. These helped me understand how data is represented, manipulated, cleaned, and prepared before it can be used by machine learning algorithms.
I also realized an important lesson:
Learning AI is not just about learning AI libraries. It is about understanding the problems those libraries help us solve.
This changed the way I approached learning. Instead of simply completing tutorials, I started focusing more on building things.

Moving From Theory to Projects

One of the biggest changes in my learning journey happened when I started working on projects.
For example, I built a predictive modeling project using machine learning algorithms such as Decision Trees and Random Forests. The project involved data preprocessing, model training, evaluation, and comparison.
The Random Forest model achieved an accuracy of approximately 95.6% and a ROC-AUC score of 0.99 on the evaluation setup I used.
More importantly, the project taught me something that tutorials often cannot:
A machine learning project is a complete pipeline, not just a model.
The model is only one part of the process.
A practical ML workflow includes:
  1. Understanding the problem
  1. Collecting and understanding the data
  1. Cleaning the data
  1. Performing exploratory data analysis
  1. Engineering useful features
  1. Splitting the data correctly
  1. Training models
  1. Evaluating performance
  1. Comparing approaches
  1. Deploying or communicating the results
This experience made me more interested in the engineering side of AI.

Why I Started Exploring Generative AI

After working with traditional machine learning, I became increasingly interested in Generative AI.
Large Language Models changed the way people interact with software. Instead of writing a program for every possible input, we can build systems that understand natural language and generate useful responses.
But simply using an LLM API is only the beginning.
I became interested in understanding what happens behind applications built around LLMs.
This led me toward topics such as:
  • Transformers
  • Large Language Models
  • Embeddings
  • Vector databases
  • Retrieval-Augmented Generation
  • LangChain
  • AI agents
  • Hugging Face
  • Local LLMs
One concept that particularly interested me was Retrieval-Augmented Generation, or RAG.

Understanding RAG

A basic LLM has knowledge contained within its training process. But many real-world applications need to work with private, updated, or domain-specific information.
RAG provides a way to connect an LLM with external knowledge.
A simplified RAG pipeline looks like this:
User Question → Retrieve Relevant Information → Provide Context to LLM → Generate Answer
For example, imagine building an AI assistant for a company's internal documentation.
Instead of expecting the model to already know every document, we can:
  1. Store the documents.
  1. Split them into useful chunks.
  1. Convert those chunks into embeddings.
  1. Store the embeddings in a vector database.
  1. Retrieve relevant chunks when a user asks a question.
  1. Give the retrieved information to the LLM.
  1. Generate an answer grounded in that information.
This opened my eyes to the fact that modern AI engineering is not simply about training models.
It is also about building reliable systems around models.

Where AWS Fits Into My Journey

As I started learning more about AI engineering, cloud computing became increasingly important.
A model running on a laptop is useful for learning, but real-world AI applications often need scalable infrastructure, storage, APIs, security, monitoring, and deployment.
This is where AWS becomes interesting for me.
The AWS ecosystem provides services and tools that can be used to build, deploy, and scale applications.
For an aspiring AI Engineer, I see cloud knowledge as an important complement to machine learning knowledge.
My goal is not simply to learn individual AWS services. I want to understand how cloud services can become part of an end-to-end AI solution.
For example, a future AI application could involve:
Frontend → API → AI application → Retrieval layer → Model → Database → Cloud infrastructure
Understanding how these components communicate is an important part of becoming an AI Engineer.

Learning Through Building

One lesson I have learned throughout this journey is that watching courses alone is not enough.
It is easy to spend hours watching videos and feel productive without actually being able to build something independently.
Projects force me to answer questions that tutorials sometimes avoid:
  • Why did my model perform poorly?
  • Is my data preprocessing correct?
  • How should I evaluate the model?
  • How do I handle errors?
  • How do I expose my model through an application?
  • How can I deploy it?
  • How do I make the system useful for someone else?
These questions have made my learning much more practical.
I have also started using GitHub to document and share my projects. This gives me a place to track my progress and gradually build a portfolio.

The Road Ahead

I am still at the beginning of my AI engineering journey.
My next focus areas are:
Machine Learning → Deep Learning → Transformers → LLMs → RAG → AI Agents → Cloud Deployment
I also want to become more comfortable with the engineering skills surrounding AI, including APIs, Git, deployment, databases, system design, and problem solving.
My long-term goal is to build AI applications that solve actual problems rather than simply demonstrating that a model can generate an output.

What I Have Learned So Far

If I could give advice to someone starting their AI journey, these would be my biggest lessons:

1. Don't try to learn everything simultaneously

AI is a huge field. Trying to learn every framework, model, and cloud service at once can quickly become overwhelming.
Build one layer at a time.

2. Projects are essential

A project exposes gaps in your knowledge much faster than passive learning.

3. Understand the “why”

Don't just memorize code.
Understand why preprocessing is required, why a particular model is appropriate, why RAG is useful, and why a particular architecture makes sense.

4. Learn cloud alongside AI

Modern AI applications need infrastructure.
Understanding cloud platforms can help bridge the gap between an ML experiment and a production application.

5. Keep building

The AI ecosystem changes incredibly quickly.
The goal cannot be to “finish learning AI.”
Instead, the goal should be to develop the ability to continuously learn, experiment, build, and improve.

Conclusion

My journey from Python and traditional machine learning toward Generative AI is still in progress.
There is a lot left to learn, but I have started to understand something important: becoming an AI Engineer is not about knowing every AI technology.
It is about being able to take a problem, understand the data, choose appropriate AI techniques, build a reliable application, and deploy it so that people can actually use it.
AWS and cloud technologies are becoming an important part of that journey for me because they provide the infrastructure needed to turn AI ideas into real applications.
I am excited to continue learning, experimenting, and building—and hopefully share more of those projects with the AWS community along the way.
The journey from learning AI to building AI has only just begun.
Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article