Sometime during my undergrad in Chemical Engineering, I figured out that I wasn't really good at that and couldn't imagine the rest of my life as a chemical engineer. I had some interest in Neuroscience but didn't really know how to transition to that.
Four years ago, During a chemical engineering research internship at a Canadian University, I stumbled upon Geoff Hinton's Lecture on deep learning. I don't remember much about what he said but I was fascinated. I started looking at more related lectures at http://videolectures.net/. I wasn't understanding much, but I was making some progress.
After I returned from the internship for another year at college, I focused entirely on learning whatever on the internet has to offer on Computer Science and AI / ML. My grades tanked but it frankly didn't matter to me. I would watch lectures, read books and read papers. After a while, I started forgetting stuff I had read. I figured I should be writing it down and having a strictly disciplined approach towards self-learning.
I started writing things down by pausing videos. It was not a good experience and my handwriting would start getting messier, the longer I wrote. I figured I should be writing in Latex. Started making notes in Org-Mode which export to a Tufte Latex Book. IMHO Tufte Latex Book templates are amazing.
I also made a lot of Anki Decks. But I don't think I ever used them more than once. There was not really a drive to do these things. I mean these were obviously important to learning something but they weren't really the things I was interested in.
The more videos I watched, the more I became convinced that the video-based platform is a bad way of learning online. They provide quick introductions but for deeper stuff, you need to read books. That's when I realized that there is a gap in the education platform offering online.
I had some thoughts about what should be the ideal system look like. I made a small script in python that teaches the user using in somewhat Socratic method, makes notes for him, generates flashcards from notes and reminds him of when to revise by calculating the Retention Score using Ebbinghaus Formula[1]. That was two years ago. I decided to pursue the idea. Quit my job a year back to concentrate full time. Right now it does a lot more. I will release soon. And bot's name is Primer. A personal tutor.
I am convinced that a bot which teaches difficult things is what the world right now. I might be wrong but it is something I can see myself working on improving the rest of my life.
I think writing all this doesn't really help you, but my point is pretty simple. Pursue things that you feel require your immediate attention or what you think you feel like doing right now. It may lead to somewhere else and which in return will take you somewhere else. But if you pursue enough randomly seemingly unconnected things, it may result in you finally have an idea which you would be convinced worth pursuing your life. Just my two cents.
P.S: I know I should be releasing early and getting feedback. I have consciously chosen to ignore that advice for a while.
Four years ago, During a chemical engineering research internship at a Canadian University, I stumbled upon Geoff Hinton's Lecture on deep learning. I don't remember much about what he said but I was fascinated. I started looking at more related lectures at http://videolectures.net/. I wasn't understanding much, but I was making some progress.
After I returned from the internship for another year at college, I focused entirely on learning whatever on the internet has to offer on Computer Science and AI / ML. My grades tanked but it frankly didn't matter to me. I would watch lectures, read books and read papers. After a while, I started forgetting stuff I had read. I figured I should be writing it down and having a strictly disciplined approach towards self-learning.
I started writing things down by pausing videos. It was not a good experience and my handwriting would start getting messier, the longer I wrote. I figured I should be writing in Latex. Started making notes in Org-Mode which export to a Tufte Latex Book. IMHO Tufte Latex Book templates are amazing.
I also made a lot of Anki Decks. But I don't think I ever used them more than once. There was not really a drive to do these things. I mean these were obviously important to learning something but they weren't really the things I was interested in.
The more videos I watched, the more I became convinced that the video-based platform is a bad way of learning online. They provide quick introductions but for deeper stuff, you need to read books. That's when I realized that there is a gap in the education platform offering online.
I had some thoughts about what should be the ideal system look like. I made a small script in python that teaches the user using in somewhat Socratic method, makes notes for him, generates flashcards from notes and reminds him of when to revise by calculating the Retention Score using Ebbinghaus Formula[1]. That was two years ago. I decided to pursue the idea. Quit my job a year back to concentrate full time. Right now it does a lot more. I will release soon. And bot's name is Primer. A personal tutor.
I am convinced that a bot which teaches difficult things is what the world right now. I might be wrong but it is something I can see myself working on improving the rest of my life.
I think writing all this doesn't really help you, but my point is pretty simple. Pursue things that you feel require your immediate attention or what you think you feel like doing right now. It may lead to somewhere else and which in return will take you somewhere else. But if you pursue enough randomly seemingly unconnected things, it may result in you finally have an idea which you would be convinced worth pursuing your life. Just my two cents.
P.S: I know I should be releasing early and getting feedback. I have consciously chosen to ignore that advice for a while.
[1]: https://en.wikipedia.org/wiki/Forgetting_curve