Optomechanical Engineer | Daniel Shteinbok

What's Nano?
6 min readNov 29, 2023

Daniel Shteinbok interned as an optomechanical engineer at Bruker Nano, an equipment manufacturer for nanoanalytics in Mountain View, California, USA. Keep reading to learn more about his role enhancing the data collected by these high-precision nanoinstruments, skills that are helpful for this job, and why calcium imaging is so fascinating to him.

How did you get to where you are today?

This particular role of calcium imaging happened quite accidentally. The project I was originally hired for was centered around the production of a coating using quantum dots. I previously worked on producing graphene at a prior co-op, so I was hired for my prior experience in producing and working with nanomaterials. However, the calcium imaging project became a higher priority and I ended up spending more time on it.

What are the main responsibilities of your job? What project(s) have you taken on?

Images and videos taken through our microscope suffer several issues stemming from optical system constraints. For example, the probe lens, sensor noise level, and optical component size and weight causes blurry and noisy images. A major problem is the form of blurring varies significantly over an image, and even across individual animals. This variation means that many traditional approaches to fixing image blurriness do not apply. My main project is developing an algorithm that can restore the quality of images taken through our microscope. Tools that I use a lot are Python with tensorflow, numpy, opencv and other somewhat standard libraries in image processing. I go through a cycle of literature review to see what approaches have already been taken, candidate strategy formulation, strategy testing, assessing problems I observe, and then addressing them.

What’s your favourite part about your job?

I like that I am given a lot of freedom in the approaches that I can take to solve the problem, and so I get to find cutting-edge papers published as of last year and implement them — this is real research! As a consequence, I get to own my solution. I was always the kind of person who liked to spend a lot of time going down Wikipedia rabbit holes, and that is pretty much what I get to do in this role (but with scientific papers). I do think that a major reason for this freedom is that I went out of my way to be self-sufficient for the cycle mentioned above, so it wouldn’t be specific to this field/job/company. I’ll explain more below.

What skills do you need for your job? What skills did you learn in your job?

Specifically for what I’m doing, a huge help was already having some experience in programming in python (and anaconda, some of the commonly used python libraries for image processing and scientific computing, etc.), using jupyter notebooks, and using linux (the company’s big server with a lot of GPUs runs Ubuntu, and should be connected to over ssh. As a consequence, a lot of my general interaction with that server is done in a terminal and knowing basic commands for things like navigating through folders, moving files, and editing files is very helpful). These skills are commonplace and probably quite necessary for almost any engineering job, and without them the learning curve would have been really steep. That being said, I wasn’t hired for having these skills; but if I hadn’t had them, just starting this project would have been really difficult and boring.

A “soft” skill that I thought was really important was to some extent the capacity to independently go through all the stages of the cycle I outlined in the Main Responsibilities section. Namely, I always pushed myself to ask and answer the question of, “what now?” In a previous co-op I had failed to do this, and ended up continuously asking what to do after each experiment. I had seen this lead to the result of the decisions with respect to the direction of the research being made by my supervisor; of course this would be the case, since I kept asking him to do it! On the other hand, I decided that I really wanted to control the direction of my research myself in this co-op position, so I interpreted all the data myself and made decisions about what to do next. In a word this would simply be called independence, and I really think that it is the key to enjoying, learning from and succeeding in any research position.

What courses had an impact?

I happened to have taken three courses that ended up being instrumental to my main project: linear algebra 2 (MATH 235), mathematical physics (PHYS 364), and statistics (NE 215). I felt that some of the stuff I learned in linear algebra 2 (most directly, singular value decomposition) was really instrumental in understanding even the first paper that I started with for this project. Mathematical physics was useful for what it taught in terms of some of the functions used in solving partial differential equations. These functions (most notably, Bessel functions in combination with commonly-known sinusoids) are used often in papers as a way to separate 2D functions in e.g. polar coordinates into a product of two functions of only one of the two variables, then represent each as a series. This turns out to be super useful for interpolating a spatially-varying blurring kernel, which is done a lot in the literature to model the problem that I was interested in. Also, a lot of the stuff that I was working with really requires a strong statistics knowledge, much more than what I have. Nonetheless, taking a statistics course allowed me to at least understand some of the words used in papers.

The theme here is that you can always go off and google the things you need to know in reading papers (this is something that I spent an unfortunate amount of time doing). But the more you know from the courses that you took, the more you will appreciate the techniques used in the papers and be able to extrapolate from the things you read in the literature review you do. Really understanding fundamental concepts in mathematics and statistics takes an initial time investment, but multiplies the efficiency of your literature review.

What’s the most surprising thing you’ve learned about: the technology you’re working on?

The most surprising thing is how far along the state of the art is in the field of calcium imaging. Even just the idea of being able to observe neurons fire in a freely-behaving mouse sounds like science fiction, but also the combination of cutting-edge technologies in different fields that are applied to this problem is shocking. For example, on the optics side there are techniques like two-photon microscopy and light sheet microscopy that take advantage of very recent advances in optics, and on the biology side people take advantage of genetic manipulation technologies like the Cre-Lox system that have only been demonstrated for this use very recently. All of these very recent advances have to come together to make much of the research being done possible. Before working here (at which point I couldn’t have even imagined that something like calcium imaging was possible), I was under the impression that all the physical principles and technologies used in research were typically old, except for the particular thing that was being examined. Likely this came from my own prior experience, where prevalent microcopy technologies in research were things like bright field and dark field microscopy, which have both been around for almost two hundred years. Even outside of microscopy, in nanotechnology engineering lab courses, the vast majority of the technology that is used (except software) is fundamentally ancient. On the contrary, it seems like there actually are a lot of research fields which really do require a whole bunch of really novel technologies, and this kind of research falls in there. I thought that was really cool.

Another thing that was surprising is how big a problem actually mapping out the neural connections is. I would have expected that observing neurons firing would be a huge technological feat, but then that simply correlating one firing with another would be essentially a piece of cake after that. This is, surprisingly, really far from the truth. Actually figuring out which neuron is connected to which other neuron and how even small parts of the brain work is one of the most exciting directions in this field of research right now.

--

--

What's Nano?

The future of medicine💊 energy🔋 electronics⚡ robotics🤖 materials 🔗