I’m interested in the intersection of text and computing. Specifically, there’s working with certain domains of text (e.g. k-12 text, patent documents), getting computers to be able to process and understand the text (at the semantic level and higher), and then trying to do it really fast, maybe in parallel (like, preferably in less time than it takes to get a coffee).
I have a pipe dream that we will at some point, hopefully within my lifetime, be able to build an intelligent thing that will help make K-12 education more personalized and better for all students at a much lower cost. Preferably though, I’d prefer to do it within Bill Gates' lifetime since maybe if I asked him nicely he might well pay for building it.
The corollary of this idea seems to be that the intelligent thing will be able to ace all the forms of standardized testing we can throw at it, since we sort of do the same thing to students. This seems to be a more tractable problem. I have a conjecture that the amount of knowledge required to ‘solve’ K-12 questions is finite. Arguably, it is the sum of the things we’re trying to squeeze into the heads of students, topped off with some common sense knowledge. Both of these things grow at a sufficiently slow pace that we could make the intelligent thing learn fast enough (i.e. manually encode that information and feed it to the thing). We shall see how that goes.
I’m also a native of Singapore.
I am attempting to grow tomatoes.
I ran a non-profit education start-up OpenLectures some time ago. It’s the first MOOC in Southeast Asia, you can still watch the videos on YouTube.
I then co-founded another edtech start-up Miao Technology, where we built a mobile app for K-12 students to get instant access to educational resources on their phone by snapping pictures of questions. I’m no longer involved in this, I got busy in college, but
Right now I’m working on some research related to automatic solving of text-based K-12 math questions. I obviously can’t put it out here yet at an incomplete state, but feel free to talk to me about it! I’m also working on building a larger text corpus for data science / machine learning purposes specific to K-12 education, if that’s another (rare) interest that you might have.
I deploy most of my longer term hobby projects on my Unraid server. This includes various data collection services (see my posts), databases, and my local music server. In addition to being intrinsically useful, it has taught me a lot about networking, hardware, VM and container management. If you are looking for a software distribution for that unused computer sitting around your house that you’d like to turn into a server, I strongly advocate for Unraid as your platform of choice!