Piyush Gupta

On brushing up math again

I love doing things that keep me on my toes. Math has always been one such thing.

With so much going on in AI lately and research papers floating around, I got curious to read some of them. I could follow the details but the moment math appeared, it went over my head and stopped making sense. It was a clear reminder that my math skills had gotten rusty.

I started with linear algebra and calculus playlists from 3Blue1Brown to get up to speed. It was entertaining but left me with knowledge without application. I got Gilbert Strang's Linear Algebra book but couldn't follow through because it was too much to get started with.

I'd seen MathAcademy multiple times on my Twitter feed but didn't take much notice. Then I discovered Justin Skycak and the insights he shares about math and learning. He's written some really great books you should check out if you're interested in math. I stumbled upon his book "The Math Academy Way," read a few chapters, and it resonated with me.

This led me to buy an MA subscription and give it a try. I took the assessment and started with the Math Foundation track. It's been more than two months since I've been practicing math problems and have completed Mathematical Foundations 1 and I'm halfway through Mathematical Foundations 2.

Finally this has been put to use

I've been enjoying doing math again. Little flex, I recently made it to the Diamond League (top 6% of learners) on their platform. What I love is their approach. It surfaces topics relevant to me, throws problems my way, and lets me struggle through them. This bottom-up approach works best for me where I'm learning on-demand and applying it simultaneously.

I've internalized that you have to mentally sweat to get better. There's no easy way out. Learning something new is a journey in humiliation. It's hard and frustrating. You have to go through the chaos to finally see clarity. If anyone's claiming to make learning fun, run.

My aim is to complete the entire Mathematical Foundations track this year and start Math for Machine Learning next year. Next year's gonna be fun.

I'm also using LLMs to dive deeper into concepts and build intuition. After experimenting with several, Gemini 2.5 Pro is the clear winner for explaining math concepts. There were times when concepts clicked—like why the denominator equals zero to find vertical asymptotes of rational functions. We're truly living in an exciting time.