What I'm working toward
Where my coursework, side projects, and reading are pointing right now — and what I'm trying to get better at next.
Most of my analog work so far has been small-signal — amplifiers, filters, bias networks. I'm deliberately moving toward power conversion, where efficiency, thermal behavior, and switching transients matter more than gain does.
The LM317 constant-current driver on my projects page is the entry point: a linear regulator is simple enough to get a first board fabricated and tested, and the line-regulation sweep it needs is the same measurement I'll reuse on everything after it. The next step is a DC-DC buck/boost converter in KiCad with a formal test report — efficiency curves across load, ripple measurement, and thermal behavior.
This is where my two tracks actually meet. A laser diode is an unforgiving load — it fails quickly under current overshoot, so driving one well is a circuit-design problem governed by device physics I'm studying separately in my graduate coursework.
I'm working from the LM317 board toward an op-amp and MOSFET precision current source, then toward a protected diode driver with proper current limiting, soft-start, and modulation capability. Getting that right means understanding both the compliance and stability of the driver and the relaxation dynamics of the device it's feeding.
My MS concentration is Physical Electronics & Electro-Optics, and my bachelor's thesis covered optical processing — photonic integrated circuits, all-optical architectures, and what happens to computing when electrical interconnect becomes the bottleneck rather than the transistor.
I'm most interested in the boundary layer: not pure photonics and not pure electronics, but the hybrid electro-optic systems where a laser, its driver, and its control loop have to be designed together. Silicon photonics and InP integration are what I'm reading toward.
During my research associate work I profiled GPU workloads on the NVIDIA Isaac stack to find inference bottlenecks, which turned into as much a power-and-thermal question as a throughput one. That experience is part of why power electronics pulled my attention.
The scaling problem in AI hardware is increasingly about energy per operation and moving data, not raw compute — which is exactly the argument for optical interconnect. It's the thread connecting the robotics work to the photonics work.
Most of my circuit work has been verified in LTspice against hand calculations. That's a good discipline, but simulation hides parasitics, layout coupling, and thermal drift. I picked up KiCad specifically to close that gap — designing a board, having it fabricated, and measuring what actually comes back.
The goal is a workflow I trust end to end: hand analysis → simulation → layout → fabrication → bench measurement → a written test report that says whether the design met spec and why.
I'm actively seeking electrical engineering internships and full-time roles — particularly in power electronics, analog design, or photonics/laser systems. I'm a U.S. citizen, so cleared and ITAR-restricted work is open to me.