My Machine Learning Journey 2024

Hi folks, this is my first blog on Hashnode! I decided to start writing here to revisit what I’ve learned and also to help others who are curious about machine learning understand what it feels like to have “AI superpowers”😁

Kicking Off My AI Journey

I began my AI journey with several short courses by Andrew Ng: AI for Everyone (2023), Generative AI for Everyone, and Prompt Engineering for ChatGPT Developers (completed by February 2024). These courses were a good starting point to determine if AI aligned with my interests and to help me understand what all the buzz was about. By the end of March 2024, after gaining a clearer picture of what AI actually entails, I switched gears and began a course called “Mathematics Specialization for Machine Learning and Data Science.” You heard that right—I started focusing on math because the foundation of AI is, at its core, math. Damien, a senior director at RBC Bank, said that “AI is just statistics on steroids.” This resonated with me and justified my decision to invest time in building a solid math foundation.

I completed the math specialization in October 2024—though I did take longer than expected, partly because I explored various Kaggle notebooks and got hands-on with data. Honestly, Kaggle wasn’t the only reason for the delay; I audited course😅By May 2024, I had completed the first course of the Machine Learning specialization, Supervised Learning, which gave me a decent understanding of how to build AI models.

When I Realized It’s All About the Data

With a grasp of classification and regression, I started some projects in June, including Iris data analysis, Credit card fraud detection, Walmart sales analysis, and Obesity-level analysis by July. To my surprise, I spent the bulk of my time dealing with data. Every Kaggle notebook followed a familiar pattern: preprocess the data, clean it, manipulate it, perform some statistical analysis, visualize the patterns, and only then think about which AI algorithm to apply. Before diving into AI, I realized how crucial data handling and exploration is.

Diving into Computer Vision

At this point, I realized that simply taking certifications and building a few projects wouldn’t necessarily give me the depth of expertise I was after. So, when my fall semester started, I enrolled in a Computer Vision course. For the first two months, we focused on traditional computer vision techniques: pixel transformations, edge detection, object detection, and interesting mathematics like Laplacian edge detection. By October, we hadn’t even touched on AI-driven methods. In the second part of the course, which ran until December, we got upto deep learning. For our course project, we worked on Red Light Violation Detection—a system where a pre-recorded file is uploaded and an AI model, with the help of OCR (Optical Character Recognition), detects which vehicle violated the red light. This project involved a lot of coding, maybe more than a traditional data scientist expects, but I’m a computer scientist🤓

What’s Next? More Courses and More Math

As I wrap up this first blog post, my next step is to finally complete the two remaining Machine Learning courses I’ve been putting off—this time, I promise not to just audit them! Moving forward, I plan to launch a blog series that dives deeper into the mathematical foundations of AI and then builds up to more advanced ML techniques. I’m super excited for this😀