Why I'm doing this

June 2026

I'm an aerospace student in the UAE. Right now I can't make a drone hold a steady hover on its own. By the end of this summer, I want to. That's the whole bet.

I've started where the math is honest: I derived the equations of motion for an inverted pendulum cart by hand, and I'm building the simulation in Python from the ground up - my own integrator, my own controllers, no black boxes I don't understand.

Four projects, escalating:

The pendulum sim - PID, then LQR, then a Kalman filter on noisy sensor data. Learn control where it's safe to be wrong.
A quaternion attitude estimator running on an STM32, fusing a real IMU and magnetometer in real time. Learn estimation on real silicon.
My own flight controller firmware on a Teensy - attitude loops, motor mixing — bench-tested on a stand before anything spins near my face.
An autonomous quadcopter that holds position and flies a waypoint mission indoors, in simulation first, then in the air.

If all four work, the hover is solved as a side effect.

Why this and not the thing everyone's chasing - I want to write code that turns into motion in the physical world. A control loop either keeps the aircraft up or it doesn't; reality grades the work, not a benchmark. That feedback is the part I can't get enough of.

It's also not abstract for me. 2026 made it concrete - the UAE got caught in the crossfire and held. I want to be in the systems layer that keeps it that way. GNC, avionics, control - that's where I think a single competent person has the most leverage, and it's where I'm pointing everything I have.

I'm writing this down for three reasons. Writing it forces me to actually understand it instead of pretending to. Future me will want the receipts - the failures logged, dated, honest. And I'm self-taught on most of this, so I want it out in the open where someone who knows more than me can tell me exactly where I'm wrong or point out my blind spots.

So: this is where I am, where I'm going, and the public record starts now.