Hello reader, I’m Fudo (well, not my real name though ). I guess you must be coming from X, and if I’m not wrong, alas, I finally made it to Google. Welcome to my blog, sigmoidit.com.
A little introduction about this blog: I started sigmoidit.com to document my AI journey and share whatever knowledge I gain along the way with this community. Why the name sigmoidit? Well, sigmoid is a widely used activation function in Machine Learning.
This article might turn out to be a long one, so please be patient with me. You won’t regret it.
My 100 Days Math Journey
Around 5–6 months ago, I began my AI journey. I jumped straight into Andrew Ng’s Machine Learning course on Coursera. He says that high school mathematics is enough to get through the course. But as soon as I started watching the lectures, I realized how weak I was in even the basics of mathematics.
Even though he carefully explains every formula he uses, I felt I wasn’t really getting it. Still, I was determined to complete both his Machine Learning and Deep Learning Specializations and fully understand the intuition behind them at any cost.
Few months ago, I stared watching Andrew NG Machine Learning Specialization course on coursera. But couldn’t understand properly due to my poor math skills (in college).
— fudo (@0xfud0) May 27, 2025
So, I started researching the best resources to learn Math for AI/ML from scratch. After digging through forums, asking GPT countless questions, and trying different materials, I settled on a few: Linear Algebra by Gilbert Strang, Calculus by Professor Leonard, and for Probability and Statistics, Harvard’s Stat 110.
I began with Linear Algebra by Professor Strang. Despite his amazing teaching style, I couldn’t understand much. I felt I should go back to the absolute basics first. So, I turned to Khan Academy’s Linear Algebra course, which has three units. I finished the first one and then moved to Calculus 1 by Professor Leonard.
To be honest, I absolutely loved Leonard’s three Calculus playlists. His teaching style is fantastic. He starts from the very basics and gradually builds up to advanced concepts, always focusing on intuition and solving problems along with theory.
Still, I struggled with some parts of Multivariable Calculus because of the 3D intuition required. Somehow, I managed to complete it, while also working on Unit 2 of Linear Algebra (Khan Academy).
By the time I finished Multivariable Calculus, I moved on to Unit 3 of Linear Algebra. It felt too abstract, and I couldn’t follow much. So, I referred to Introduction to Linear Algebra by Gilbert Strang, and surprisingly, his explanations were easier to digest. At that point, I stopped Unit 3 on Khan Academy and kept Strang’s 18.06 course on my list.
Next, I studied Probability and Statistics on Khan Academy and completed it without major difficulties. Just like I tried with 18.06, I also attempted Harvard’s Stat 110 but found it too advanced, so I went back to basics with Khan Academy first.
Since I still had gaps in Linear Algebra, I decided to bridge them with the Mathematics for Machine Learning course on Coursera (Imperial College London) before restarting Andrew Ng’s Machine Learning Specialization. I completed it, but I still felt a bit unsatisfied with my Linear Algebra preparation. May be, only Strang can help me with that. However, gained enough Linear Algebra knowledge that required for ML specialization course.
Meanwhile, I also explored Python data science micro-courses on Kaggle, watched 3Blue1Brown’s Essence of Linear Algebra and Essence of Calculus series, and learned foundational statistics from StatQuest.
Also, I widely used these two resources for intuition :
- 2D graphing – Geogebra 2D graphing
- 3D graphing – math3d.org
So, that’s my 100-day journey of learning math for AI from scratch. I spent 3-4 hours daily, and on weekends or holidays, a minimum of 8 hours. Weekdays were mostly for studying, while weekends were a mix of studying and problem-solving practice.
This is my journey so far.
My math practice is not enough that it becomes my muscle memory, but I practiced enough to get started with basic AI foundational courses like Andrew Ng, which I struggled to understand before.
Will write a detailed blog on this journey soon. pic.twitter.com/L8FoJ44OLW
— fudo (@0xfud0) August 30, 2025
(Spent remaining 10 days on lots of calculus and probability practice + StatQuest playlists)
Now, I’m back to Andrew Ng’s Machine Learning course on Coursera, the same one that inspired me to begin this 100-day math journey. This time, I’m thoroughly enjoying it because I finally understand the intuition and derivations behind the algorithms with whatever minimal Math I learned. When you truly get the math and intuition, implementation feels so much easier. I’m loving every bit of it.
But recently, I realized that a beginner could probably learn faster and with less effort just by choosing the right resources. Kindly, refer this article on my AI roadmap and resources My AI/ML Roadmap for beginners To Advanced
Thank you for reading, and keep visiting my blog!
Leave a Reply