Tesla

Since joining Tesla as a dimensional engineer in August of 2020, I have developed and validated locating strategies for a variety of suspension components – as well as HVAC and electrical components. These strategies support both the manufacturability and the performance of each system, ensuring that parts can be assembled in volume, and that when they are, they support the performance objective of the system.

ASML

I began my work for ASML in 2014. At that time, my job was to the design of the frame for a precision optical sensor. Known as ORION, this sensor helps ASML's lithography tools to align layers of microchips to within nanometers of each other. In order to achieve this precision, the frame itself has to be stable at the level of picometers from measurement to measurement. It has to be robust to asymmetrical thermal loads and dynamic input present within our machines. And it requires small tolerances on difficult-to-machine surfaces in order to accurately locate the optical subassemblies within.

I worked on other, smaller parts during this time as well – including dynamically isolated covers, an athermal, flexured strut, and an athermal measurement bracket. While mechanically simpler, each of these parts presented unique challenges.

In 2017 I changed roles. Much of the mechanical design had already been completed, and there was a hole to fill in manufacturing. Here, my team and I developed procedures to build and align the sensor. One unique challenge we faced involved the bonding of optics to metal substrates: due to the stability requirements of the module, our bonding process had to be developed such that no stress was introduced to the bonded joint. Further, adhesives had to be tested, and optics had to be positioned accurately throughout the curing process. This led to the development of some novel tooling and processes for which my team and I had extensive input.

In 2018 my team grew to include supply chain engineering and equipment engineering. Interestingly, at this point, much of my work returned to the frame. While bringing on a new supplier, it became evident that some challenges needed to be addressed. We used SPC extensively to identify several issues, and helped our suppliers to solve these. Often, our solutions drew on the same principles of precision engineering implemented in the design of the frame itself.

Finally, in August of 2019, I transitioned to taking ownership of several test stands for manufacturing and qualifying the next generation of the ORION sensor. In this role, I worked across disciplines – including optical, mechanical, electrical, and software development – helping to resolve issues and make key decisions regarding the stands’ design and construction.

Weaver Wind Energy

I was hired by Weaver Wind Energy to perform analysis on a 5 kW wind turbine in order to certify it to American Wind Energy Association (AWEA) standards. In New York, turbines certified according to AWEA standards allow consumers to credit the energy they produce at better-than-market prices. The standards largely concern safety and reliability, and FEA and hand calculations are employed to ensure robustness to extreme weather events.

Of course, at a small company, one wears many hats. In addition to performing analysis, I also designed several components, performed bench and field testing, developed marketing material, and of course, built and installed wind turbines.

Unfortunately, Weaver Wind didn’t last. But I learned a lot in my time there, and I continue to employ that knowledge today.

Final Project: NOTAR Helicopter Boom

My final project was to design the 2D geometry for the boom of a helicopter with no tail rotor (NOTAR). The counter-rotating force came from a combination of rotor wash acting on the boom as well as jet actuation provided from within it. In designing the geometry, the idea was to create stable operating conditions with minimal control needed from jet actuation. To accomplish this, I chose an airfoil shape for the boom, mounted at such an angle of attack as to be balanced in normal operation. This project involved heavy iteration on multiple airfoil geometries in Fluent CFD, as well as Matlab modeling to interpolate around the CFD outputs.

Sumobot

Every year, Cornell hosts a tournament between project teams in its mechatronics class. The contest pits teams’ robots against each other; the goal is to push your competition’s robot out of the ring before it does the same to you. Though teams devised many strategies to knock out other robots, fortune typically favored the simplest designs. We fitted our robot with two (independently sampled) sonars and a rat trap to flip and immobilize other robots. The trap didn’t work and our robot didn’t do so well, but it was a great way to learn about circuits and embedded systems.

Here’s some video footage of the same event, a few years later, featuring a few colleagues: https://www.youtube.com/watch?v=z0Ozafq9Rlo.

Water Pump

This was one of my favorite school projects. The challenge was to lift 1 liter of water to a height of 1 meter in less than 1 minute. To complicate things, power was provided via wind from a large fan. This wind was captured by a turbine and driven to the input shaft by a belt. In order to make efficient use of this power source, we had to first characterize it and then design our pump around it. Therefore, much of this project involved building a Matlab model and using it to define the most optimal cylinder dimensions and valve timing. With these parameters, we could then create a CAD model with the goal of making the pump as inexpensive as possible. Our time in the machine shop counted too!

Bike Crank

Bike cranks are designed to be light, but they also need to be strong enough to stand up to thousands upon thousands of load cycles. For this project, we were tasked with designing a bike crank to be as light as possible and still be resilient against fatigue. In order to determine loading conditions, we first took data from strain gauges affixed to the shaft of a stationary bike. This allowed us to measure torque, and therefore define our inputs. Next we designed a model and checked it against these inputs in FEA. We then iterated on materials and geometry, until we came away with something near-optimal.