Supported early-stage medical device R&D through physiological signal processing and computational modeling work. My projects involved processing PPG data, building analysis workflows, and applying statistical methods to characterize signal behavior across experimental conditions.

I developed modeling tools for vascular geometry studies and gained experience with biomedical simulation (digital twin), parameterized geometry generation, and research-oriented engineering workflows.

Worked on the DUALTO™ Surgical Robotics Energy System team, supporting calibration fixture testing, verification, and technical documentation for internal engineering use. I created regulated technical documentation such as operating procedures, software validation documentation, and test case data to support fixture verification. These are formally reviewed and signed through the team's internal document-control process. I conducted failure analysis on the fixture module, successfully identifying issues and debugging the module.

Alongside the fixture work, I had a great opportunity to design a 4-layer ATmega32u4 development board. As my very first dev board design experience, I learned the whole board design workflow in KiCad. I constructed a design plan within a set time frame, including block diagram development, schematic design, PCB layout, component research and ordering, assembly, and bootloader flashing. This board is detailed further in the Projects section below → Demo video is attached.

An end-to-end PCB design project in KiCad, taken from block diagram to a working, bootloader-flashed board.

The board is a 4-layer, 2000 × 700 mil ATmega32u4 (7×7 VQFN) dev board. Core blocks: USB-C for power and serial, a voltage regulator feeding the MCU rail, an I²C level shifter into a proximity sensor, NeoPixel LEDs, an oscillator, and a reset button. USB-C was chosen so the board could be powered, programmed, and serial-debugged over a single connector, with the ISP pins (SCK, PDO, PDI, RESET) mapped explicitly before layout.

Layout went through two full revisions after design review, tightening placement and routing around the MCU. From there: wrote the assembly and test plans, ordered parts, hand-assembled the board, inspected the QFN joints under a microscope, verified continuity, and flashed the bootloader.

The board runs a closed-loop demo: the proximity sensor streams distance over I²C, and the firmware maps that reading to NeoPixel brightness in real time.

One of my first breadboarding projects, and the one where the fundamentals clicked. An IMU reads tilt angle on three axes; tilting right sends a VOLUME_INCREMENT and tilting left a VOLUME_DECREMENT, delivered to a phone over Bluetooth as HID commands so the setup appears as a standard Human Interface Device.

Building it taught me the fundamentals I now rely on: I²C between the IMU and the nRF52840, BLE HID over the Adafruit ItsyBitsy nRF52840 Express, and getting two chips to share timing so their data streams stay synchronized. Written in CircuitPython.