Computer Engineer & Founder
Building an AI chip company before 30.
I built my first computer at 15. Studied Computer Engineering at Waterloo. Now building Quettos and designing inference chips for agent workflows.
View selected work01 / PROJECTS
SELECTED PROJECTS / 06A left-turn collision warning system for novice drivers. Our team built a vehicle-mounted system that detects oncoming traffic and assesses threats, with quantized YOLO11 running entirely on a Jetson Orin Nano.
I built and optimized the embedded inference pipeline: compiling OpenCV with CUDA, converting the model with TensorRT, and tuning live video processing end to end.
I trained an RL driving agent to drift and change lanes in pursuit of the shortest path. A course project exploring how learned behavior translates into motion on a simulated track.
The demo shows the driving behavior in action: corner cutting, drifting, and lane changes as the agent navigates the course.
FURTHER DOWN THE STACK
Explore GitHubWeight-streaming LLM accelerator in SystemVerilog with a custom instruction set and quantization compiler. Generates text with Qwen2.5 in cycle-accurate RTL simulation.
04 / ARCHITECTURE5-stage pipelined RISC-V (RV32I) processor in Verilog with data forwarding and hazard detection.
05 / SYSTEMSReal-time operating system built from scratch for STM32, with preemptive deadline scheduling, a buddy memory allocator, and ARM hardware context switching.
06 / OPTIMIZATIONREINFORCE policy gradient agent for FPGA LUT optimization, minimizing circuit area and depth through learned merge strategies.
02 / THE NEXT CHAPTER
IN PROGRESSI'm building Quettos: inference chips designed around how agents actually work. Hardware shaped by reasoning, tool use, and memory.
Explore Quettos
03 / JOURNEY
I built my first computer at 15. I've been going further down the stack ever since.
Follow the next chapterPutting together my first computer turned curiosity into something I could actually switch on.
Computer Engineering: building compilers and operating systems, designing a CPU, and connecting software to the hardware underneath.
Running TurnSight AI on embedded hardware for my capstone. Training an RL car to drift and change lanes. Making the theory do something.
Taking everything I've learned toward inference chips for agent workflows—and an AI chip company before 30.
04 / CONTACT
Interested in AI hardware, embedded systems, or working together?
Connect with me on LinkedIn.
Connect on LinkedIn
