Kelly Ann Pawlak joined Atom Computing in 2022 as a Quantum Applications Engineer. She earned her PhD in condensed matter physics from University of California- Santa Barbara in early 2020.
Her dissertation titled, “New Directions in Correlated Electronic Systems” explored the limits of cutting-edge theoretical, numerical, and experimental techniques in elucidating the quantum properties of electrons in many exotic materials.
Graduating during the pandemic, she joined UCSB physics as academic personnel to help with emergency course transitions. She built, ran, and trained graduate students for online DIY experimental physics courses. She also wrote software that delivered free online manuals to students – in place of the expensive published versions – that is still in use by multiple science departments at UCSB.
Returning to research in hopes of helping to tackle some of the hardest problems in condensed matter physics, Kelly started working in the quantum computing industry in early 2022 as a quantum applications scientist before moving to her current role.
We talked with Kelly about her role at Atom Computing, about quantum computing applications, and how she got into this field.
Tell us about your role as a Quantum Applications Engineer at Atom Computing. What does this job involve? What projects are you working on?
Quantum Applications Engineers at Atom Computing have a broad focus, ranging from benchmarking and characterization of our neutral atom devices to error correction and mitigation, to algorithm and use-case development. I mainly work on the latter.
What does it mean to do algorithm and use-case development? At times, it means solving fun puzzles with three pieces: I have a quantum computer with these strengths and these weaknesses, I have knowledge of potential use-cases that are important, and I have a good understanding of what algorithms have been worked on and which have been fruitful. My job is to synthesize this information and push the right boundaries, producing unique ideas that take advantage of our very unique hardware to approach these meaningful use-cases. Luckily, Atom’s hardware has a lot of strengths, so this is usually pretty fun!
Other times, algorithm and use-case development means rolling up your sleeves and doing pages of math by hand or developing software – both things that I also really enjoy. I find that my position has a great balance of technical activities, and I always have something to work on and people to talk to.
What am I currently working on? Currently I’m working on algorithms in optimization, quantum machine learning and quantum simulation. Some of these projects are with partners or research collaborations, which is incredibly fun – nothing beats the feeling of working on a problem with people who are deeply invested in the results.
There is a lot of discussion about “killer applications” for quantum computing? What do you think are the most promising near- and long-term applications for quantum computing?
I love this question, and I have a vision for the near term and why anyone should care – so bear with me. As background, I worked in a field of quantum physics (condensed matter physics) that absolutely relied on quantum experiments to make scientific progress. By experiments, I mean things like epithelial crystal growth or “high temperature direct reactions”, where labs had to fabricate tens or even hundreds of samples of quantum materials for each “clean sample” good enough for data collection – which, by the way, sometimes destroys the sample! Measurements are taken and compared with theoretical expectations, allowing us to fine-tune our understanding of what is going on quantum mechanically, as well as improve the fabrication process, hopefully moving towards desired properties.
Ultimately, this means that if we want to test how observation Y depends on parameter X in material Z, you must wait months for samples to be fabricated for each X you need data for, and hope that the results are conclusive. The experimental loop in many instances is very slow, pushing scientists to either propose theories without sufficient data – leading to research rabbit-holes – or to supplement their theories with computer simulations.
The problem with computer simulations of quantum physics is that, while they do help in some cases, they are very limited. And, currently, it’s not even a problem with computational speed that a GPU or HPC can help with: Storing the quantum state of just 52 atoms that only have a spin degree of freedom (up or down) requires over 500 Terabytes of RAM. Fifty-three atoms requires over a Petabyte, and each atom doubles the amount of RAM we need. And while there are classical methods for shaving that down significantly, such as “tensor network” approaches, these require unrealistic models, severely truncating the space of states or changing the material geometry. In addition to this, very few people on earth – maybe 15 – know how to build such a simulation and configure it to give reliable results for any problems of real interest.
On the flip side, to store a state of 52 atoms, it takes 52 qubits. For 53 atoms, it takes 53 qubits. And so forth. And as gates get better, qubit counts get larger, and coherence times improve, we are going to see a shift in R&D practices starting at the academic level. We are already seeing it with the youngest generations of condensed matter and materials physics researchers – many have already embraced this technology, including some experimental labs. It is so much easier to generate, manipulate and collect quantum data on a quantum computer, and use that to improve theories and test hypotheses than it is to create a clean crystal and perform neutron scattering, or even build a reliable classical computer simulation. I think that quantum computing will find its first clear practical applications as an intermediate, highly accessible, step between computer simulations and experiments very soon, aiding in impactful research with a plethora of industry applications down the pipeline.
Speculating here, I think that quantum computers are on the cusp of providing research-grade quantum data. We’ve already seen low-control quantum simulators – as early as 2017 – find brand-new physics. I think that there will be many more surprises – new understandings of quantum correlations, and how deep corners of Hilbert Space conspire together to create unexpected phenomena. I think there will be a quantum engineering renaissance driven by ever improving quantum computing devices. I think we will have new ideas for materials, quantum devices, communication protocols, that will all be discovered using the quantum computing platforms we build in the next few years. I think that quantum computing has a real shot at being the foundational technology of new, old, and unimagined industries.
For the long-term, it is really hard to speculate on which industries will benefit from large, fault-tolerant quantum computers. Beyond the scientific uses I’ve already mentioned, I think there is a lot of excellent work on applications in optimization, machine learning and chemistry. Some recent papers have found evidence of fault-tolerant super-polynomial speed-ups in some of these areas. An obstacle for research, however, is that it is hard to study quantum algorithms without being able to run them on a large quantum computer. It’s akin to asking a computer scientist to formally work out how a neural net will behave on a training set with x, y, and z parameters, without letting them actually run the program. I think that once we have some quantum computers that are regularly demonstrating some general version of quantum advantage, we will have a lot more to say about the long-term use cases. I am optimistic!
Is there a particular quantum computing application you enjoy working on the most?
If you couldn’t tell by my previous responses already, I really love quantum simulation! Being so close to my PhD studies, I feel like this is the topic I can be the most creative with. Other than that, I’ve been enjoying working on optimization because of the intense industry interest. It’s always a pleasure to work with others who are laser-focused on solving specific sets of problems and helping them reach their goals.
Why did you join Atom Computing? What excites you most about working here?
I joined Atom because of the company’s scientific culture. The company is filled with scientists who understand the field at a deep technical level. A lot of them come from other quantum computing companies, too, bringing that experience with them. I also joined because it was very clear to me that Atom cares for its employees. I feel so welcome and valued here.
Working here is very dynamic, and you are always working on something new. The company also takes feedback seriously. Working here means that you have a voice in Aton’s direction, and it is very exciting to see your ideas incorporated into key decisions. It is so rare to have a company culture where individual scientists can make a company-wide impact.
How did you get into quantum computing? What do you enjoy most about this field?
My graduate school was a major player in quantum hardware, home to both StationQ and the Martinis Lab. There were also theory programs on the topic through StationQ and the KITP. Hilariously, I thought quantum computing was “boring” and never showed up to a single event or took a course. My opinion changed when Chris Monroe gave a talk demonstrating how an early prototype of an ion trap quantum computer could simulate some special physics models. It blew my mind, and by the end of my PhD, I decided that the way forward in quantum sciences would depend on various forms of quantum computing in the future.
I really enjoy the diversity of thought in this field, especially with more people migrating from more traditional branches of physics, chemistry, and computer science every day. Once you get past the cautious pessimism of many over-hyped applications, you can work creatively on problems. It’s a very exciting time and there is a lot to do.
What advice do you have for people who want to work in quantum computing?
I will limit my advice for people with interest in applications, since I haven’t worked on hardware.
If you have a PhD or similar in a hard-science field: really start looking at applications that intersect with your specialty and start writing some simple algorithms using a high-level language like Qiskit. There are tons of tutorials and forums out there. If you have a specific topic of interest, like error correction, just start learning and writing code. You’ll learn a lot faster than you expect. Put a project or two on GitHub and apply to some jobs. Use the feedback you get to keep improving!
If you have a software background: quantum computing companies all need software engineers for control systems, full-stack web development, compilers, etc. Look around for some roles that fit your expertise and brush up on physics and linear algebra. When you work at a quantum company, you typically can cross collaborate with other teams, so you could certainly pick up some applications-layer skills.
If you are an undergraduate: Take your quantum classes seriously and do as many quantum internships as possible. If you can’t find a quantum internship, do a software internship. It’s possible to get a job out of undergraduate in quantum computing, though many students decide to continue their education into a master’s degree or PhD. If you decide to go this route, make sure you choose a school with a quantum engineering or quantum computing program that aligns with your interests. Not every graduate school has one!
And finally, for anyone else: what I love about this field is the nonlinear paths many of us have. Pursue what projects you have time for and jump into tutorials. Many people think the Qiskit tutorials are a great place to start. Go to meet-ups, join forums, and talk to people in the industry to get an idea of how to chart your own path!