Embedded, FPGA & Board-Level Systems
Board-level engineering — Verilog on the Basys 3, bare-metal ATmega2560 work, a custom PCB with production Gerbers, and CMOS VLSI design.
Role — Design, firmware, and verification across the listed projects
- AI systems eventually meet a voltage rail. This body of work is about being competent at the layer where software constraints become physical ones.
- Embedded / board-level systems portfolio
- Physical-AI ambitions without hardware literacy produce systems that can't be debugged below the API. This layer keeps the rest of the stack honest.
- Individual coursework and personal projects across the computer-engineering curriculum and three faculty research labs.
Overview
A collection entry for the hardware layer under the AI stack: six Verilog systems on the Digilent Basys 3 (soft-core CPU, PicoBlaze + OLED, PS/2-to-UART), six ATmega2560 projects including a custom Arduino-Mega-class PCB with production Gerbers and a bare-metal 40 kHz ADC sampler, transistor-level CMOS designs in 0.6 µm SCMOS, and a self-balancing robot prototype running PID control with a custom integration PCB in progress. This is why the deployment work treats latency, timing, and power as real numbers instead of abstractions.
Per project: sensors and IO into a microcontroller or FPGA, firmware or HDL in the middle, actuation and bench validation on the other side — documented per repository.
- Sensors / IO
- Microcontroller / FPGA
- Firmware & HDL
- Actuation
- Bench validation
Contributions
- Six self-contained Verilog systems on the Basys 3 (Artix-7): PicoBlaze + PmodOLED, XADC-to-Arduino SPI bridge, PS/2-keyboard-to-UART stack, soft-core CPU, FSMs, LCD driver.
- Six ATmega2560 projects including a custom Arduino-Mega-class PCB with production Gerbers, a bare-metal 40 kHz ADC sampler, and a hand-built IR link-layer protocol.
- Transistor-level CMOS design in 0.6 µm SCMOS: 8-bit ALU, R-2R DAC + flash ADC, transmission-gate MUX, full static-CMOS gate library.
- Self-balancing robot prototype: PID stabilization, with a custom PCB (upgraded MCU, onboard drivers, integrated IMU) in design.
Evidence & evaluation
Evidence
Open-source repositories
attachedThree documented repos: FPGA portfolio, ATmega2560 projects, CMOS VLSI designs.
Bench test results
pendingScope traces, validation runs, or demo clips for the strongest 1–2 pieces.
Metrics
Not yet measured
Only the ones actually measured on the bench.
Limitations
- This is a collection entry — its credibility improves sharply once one project is documented end-to-end with bench evidence.
Lessons & tradeoffs
- Hardware debugging builds the habit that carries into ML systems work: distrust the layer you can't observe.
- [Add one concrete bench story.]to fill