also known as primitive
I’m trying to decentralize tech, from the chip up.
Right now that means AI compute. I study electrical and computer engineering at Vanderbilt, and I got here through a processor I couldn’t afford.
Most of what I do is abstraction. My days end at local rock clubs and jazz.
Whoever owns compute decides who gets to use AI. I want more people on the owning side.
This year my team and I are studying how AI compute is accessed and used, by talking to the people who run it.
See decentralization. Underneath it sit chips, photonics and quantum.
For me this is a plain question: who controls the systems we rely on? I ask it at the layers people rarely look at, like low-level classical, photonic and quantum hardware.
I learn from Joshua Tan and the Decentralization Research Center.
It started somewhere else: decolonization.
For my first 16 years my hair was buzzed to a few millimeters, because short meant civilized. At boarding school I let it grow. Afro, braids, cornrows, coils. I’ve been decolonizing the world around me since, starting with myself.
In tech I ask the same thing about computers, and about who gets to do research.
My uncle gave me Morris Mano’s Digital Design for my birthday. I wanted to build what I learned, but I couldn’t afford hundreds of ICs and breadboards.
So I built an 8-bit processor gate by gate in a free simulator. I’ve been working down the stack since.
I work on estimating OTOCs on circuits of up to 38 qubits with statevector, MPS and Pauli propagation simulators.
For a satellite that does quantum key distribution, I work on FPGA timing at the Vanderbilt Quantum Initiative. That work runs on single photons, so it leans on photonics.
I lead a reprogrammable UART/SPI/I2C engine for Tiny Tapeout. Its UART fits in 89 standard cells.
It started with a processor I couldn’t afford. Next, we plan to put our satellite’s optics on a photonic chip.
MachDesigner is a silicon photonic simulator I built. Its two independent engines agree within 1%.
My days end at local rock clubs and jazz corners. It’s where I stop doing abstraction.
Africa has 17% of the world’s people and produces about 1% of its research. ARPL trains African high school students to publish theirs. I’m its Chief Program Officer.
Find the pattern in a system, keep the variables that matter, and drop the rest. Then repeat until you can work with it.
I once knew almost nothing about OTOCs, so my team turned each simulator into a function of time and cost and optimized over them. It’s how I learn AI compute, chips and quantum.
A primitive is the smallest operation a system is built from. It’s also my nickname.