Jan. 2021 - Apr. 20214 min read

A-Star Has No Idea You're Bluffing

Building AI opponents for a tactical card game taught me that classic search (A*, BFS, DFS) is great at finding paths and terrible at reading a player. The fix wasn't a smarter algorithm — it was a heuristic that pretended to.

C++Game AIAlgorithms

Watch the demo Source code

Tactical games taught me the difference between path-finding and decision-making the hard way. The AI opponent could beat a human at board control. It could not handle a human making a deliberately suboptimal move to bait it into one.

The fix wasn't to plug in a neural net (we didn't have the compute or the time). It was to write a heuristic that overweighted human risk-taking and undermined the AI's overconfidence — close enough to feel like an opponent, far enough from optimal to stay beatable.

Related project

Multiplayer Game with AI Mechanics

View the project

Wiring a Glove to a 3D Hand

Quaternions, IMUs, and a Wireless C++ App That Doesn't Lie About Drift

Reading IMU data over wireless and turning it into a stable 3D rotation in OpenGL is mostly an exercise in fighting drift honestly. The hardware lies; the math tells you when.