Self-Balancing PID Robot

A two-wheel robot that stays upright on its own. An IMU measures how far it's tipping and a PID loop drives the motors to catch the fall — hundreds of corrections a second.

1. CAD & chassis — full mechanical design finished and the frame built.
2. Electronics wired — motors, driver, IMU, and microcontroller all connected.
3. Tune the PID loop — dialing in the gains so it balances and rejects pushes.

8-Bit Computer

A working computer built from scratch on breadboards, one logic chip at a time — clock, registers, ALU, RAM, and control logic, following the classic SAP-1 architecture.

1. Research & plan the build — study the SAP-1 architecture, source the parts, and map out every module.
2. Clock module — adjustable speed with single-step for debugging.
3. Registers & ALU — A and B registers on the bus, plus add and subtract.
4. RAM & program counter — memory with an address register, and instruction sequencing.
5. Output & control logic — the 7-segment display, instruction register, and microcode.
6. Program it — write machine-code programs, load them into RAM, and run them.

Latest · June 2026 — Registers moving data cleanly across the bus; building the ALU next.

Simple RL Robot

A small servo robot that learns to move on its own. I train a reinforcement-learning policy in simulation, then deploy it straight to the hardware — no hand-coded gait.

1. Research & learning RL — getting the fundamentals of reinforcement learning down before building.
2. Build the robot — servos, frame, and electronics assembled.
3. Match it in sim — recreate the robot in a MuJoCo simulation.
4. Train the policy — run PPO until it moves reliably in simulation.
5. Sim-to-real — deploy the trained policy onto the physical robot.

Latest · June 2026 — Learning the fundamentals of reinforcement learning before the build.