taykim.hashnode.devRaftDistributed systems should be fault tolerant - we want the system to stay available and continue operating "correctly" even when some nodes fail. One way to do this is to run replicas, so that if the 1d ago·12 min read
taykim.hashnode.devOrder and Consistency in Distributed SystemsThe Ordering Problem: From Physical to Logical Time When reasoning about distributed systems, it is helpful to find the order of events. Given this order, we can verify whether the sequence is “correcFeb 9·9 min read
taykim.hashnode.devHarmonic Coordinates for Character ArticulationSome time ago, I took a course on Geometric Modeling. Recently, I was going through my old notes and project, and I thought it would be nice to write a short post about what I built before I completely forget the details. This post is a quick look ba...Jan 30·7 min read
taykim.hashnode.devRooflines on NVIDIA GPUsThis semester, I took a course called Cloud and Machine Learning. One of the core topics covered was performance profiling, and we had a project where we used NVIDIA’s profiling tools to collect GPU metrics while training a deep neural network. In th...Jan 2·6 min read
taykim.hashnode.devOS Family TreeOperating systems was one of my favorite classes at school, but I realized I didn’t know much about them in the real world — what distros are out there, how they differ, and why you’d pick one over another. Whenever I use the cloud or spin up VMs, I ...Dec 24, 2025·6 min read
