Avital Szulc joined the Convergent team five years ago and has played a central role in developing PEAK IQ, Convergent’s energy storage intelligence.
Avital Szulc joined the Convergent team five years ago and has played a central role in developing PEAK IQ, Convergent’s energy storage intelligence.
Now Vice President, Data Science and Machine Learning, Avital leads the development and optimization of PEAK IQ to ensure energy is stored and dispatched at the most strategic times. PEAK IQ uses the latest artificial intelligence (AI) and machine learning technologies to seamlessly maximize asset performance.
When she’s not busy training algorithms or catching up on the latest research in the field, you can find this native New Yorker in Central Park enjoying everything the city has to offer with her family.
How did you get into the energy industry and why did you decide to work at Convergent?
I was working at a Philly-based battery storage company the summer after my junior year of college. While there, I had a close colleague who talked with me about the energy industry and was always telling me how great renewables were—that got me interested.
I was invited to join a small startup where I worked for three and a half years. Unfortunately, as these things sometimes go, the startup was not growing at the pace that I was hoping it would, so I started looking for a new role. At the time, Convergent was only a dozen people. I loved the ability to impact the team directly while also being pushed to learn quickly. I came to the team as a product manager and started building out software—something that could only really happen at a company of that size. It was a dynamic learning environment.
Can you tell us more about your background?
I received a Bachelor of Science in Engineering from the University of Pennsylvania in Chemical Engineering and graduated with a minor in Economics and Energy and Sustainability. After I built a good working knowledge at Convergent, I went back to school at Cornell Tech to earn a Master’s in Operations Research. Having the opportunity to partner practical experience with an academic setting was great.
Fortunately, there is also a lot that you can teach yourself in the internet age and I spend a lot of time looking into the theory behind data science and reading academic papers.
Where do you see data science in the energy field going in the next five years?
Over the next five years, as the complexity of the challenges increase the hardware will also get more sophisticated. We are always going to be working to find the best solutions.
Today, there are many different components that work in tandem across energy storage and solar-plus-storage project. There are local regulations, opportunities to value-stack across markets, and different hardware systems to consider and weigh. The goal is always to optimize how these systems work together to achieve our goals. One of the biggest ongoing challenges—and opportunities—as a data scientist is understanding how everything works together.
How would you describe the role between AI/Machine Learning and the electricity grid?
The “intelligent” component of Artificial Intelligence comes into play when we need to determine how we are going to use the energy that gets produced. Batteries don’t produce energy, they store it. Having an intelligent battery that is able to weigh critical variables and dispatch the stored energy at the most strategic times is our goal.
A useful battery storage system lets customers participate in markets that they might not have otherwise while saving them large sums of money. Intelligent dispatch unlocks doors and, in many ways, can bring relief to the aging infrastructure of the electricity grid.
How does PEAK IQ set us apart?
PEAK IQ has been in development for over a decade and is one of the world’s longest-running, most trusted energy storage intelligence platforms. The primary benefit of PEAK IQ is its unparalleled operating experience, performance metrics, and algorithms.
What is the Difference between Machine Learning and Linear Algorithms?
Machine learning algorithms can help you predict future behavior based on historical data. Linear programs are very powerful and they are used when you’re trying to maximize or minimize an objective based on a series of constraints. They can be used to co-optimize for multiple use cases. However, they’re not predictive. Both Machine learning and linear algorithms are strong tools, but they have different uses.
Walk us through your typical day as a data scientist at Convergent.
A lot of my time is spent experimenting with different kinds of models. For example, when entering a new market, the first step is to gather on the ground understanding. From there, I pick the classes of algorithms, select the types of data, train the model, and then test. Working with machine learning is an inherently iterative process.
What excites you most about the work you are doing?
I love the research component and learning about new models and algorithms and how to apply them in different ways.
There is a huge shift in the conversation around energy storage—both for environmental and financial reasons. Helping to develop the algorithms that pull everything together in such a rapidly growing industry is energizing … pun intended.
What do you wish more people knew about PEAK IQ?
Algorithms are only as good as the data they are based on.
How do you spend your time outside of work?
I have a 15-month old son named Leo! We live close to Central Park and like to spend our time outside and with family. We love traveling, going hiking on the weekends and going to the climbing gym.
What advice would you give to a young data scientist / your younger self?
Take more coding courses in school!
—
Is Energy Storage Right for You?
If you’re interested in learning more about how energy storage can benefit your business or community, please contact us today.
—
Want to work at Convergent?
Convergent is hiring. See our open positions here.
