Tan Mit
Notes on programming, how a system gets put together, and the things outside work that I want to think through.
Opening the post list…
Technology, AI, programming, and work2 min read
These three ideas show up everywhere in algorithm problems. The point is not to memorize the template. It is to see what information you have to keep, and why the answer stays correct as the window moves.
Technology, AI, programming, and work2 min read
A team can merge a lot of pull requests and still release slowly, because each person is holding a different picture of the system. Design and planning are how those pictures become one.
Technology, AI, programming, and work2 min read
Not every AI feature should be a chat box. When the product has to classify and decide quickly, a decisions API and a rule engine are often faster, cheaper, and easier to trust than a model that talks.
Technology, AI, programming, and work2 min read
A coding agent can finish a feature before the coffee cools. The debt shows up when someone has to change it. Speed is real. So is the cost of code nobody can explain.
Technology, AI, programming, and work1 min read
Children like playing games. Scratch lets them make one: a character, a puzzle, and the bugs they have to think their way out of. The point is the making, not a perfect game.
Technology, AI, programming, and work2 min read
A model can draft an API, a fix, or a small app from a description. If typing speed stops being the advantage, the developer is paid for judgment: what to build, what to refuse, and what has to stay true.
