I built and open-sourced a flying drone from scratch in my university dorm during first-year engineering.
Not assembled. Built. Everything. Including firmware.
I designed the frame, 3D printed the body, wired the electronics, and wrote the flight firmware from zero.
I took OSDN, a brand-new linear-attention model that learns to tune its own memory updates as it reads (think AdaGrad for the architectures trying to replace the transformer), rebuilt it from scratch in pure C++ with my own autograd engine, and ran it on a $4 microcontroller to predict hypoglycemia 60 minutes before it hits.
No PyTorch. No JAX. No TensorFlow. No ML library at all. Straight C++ standard library.
It runs on an ESP32-S3 in 20% of its RAM.And the deploy kernel is proven: three independent implementations of the math agree to 8 decimal places, checked on every commit. Next step is leave-one-patient-out cross-validation. https://t.co/KdNxz34tFG
I built and open-sourced a flying drone from scratch in my university dorm during first-year engineering.
Not assembled. Built. Everything. Including firmware.
I designed the frame, 3D printed the body, wired the electronics, and wrote the flight firmware from zero.
The hard part wasn’t making the motors spin. It was building the flight stack: sensor fusion, PID control, motor mixing, wireless communication, telemetry, calibration, and enough tuning to make it actually fly.