Stephanie Klaskin joined Andesite in March 2024 as a Data Scientist working on our engineering team. Her passion for research design and a knack for problem solving uniquely position her to explore new capabilities and push our product forward.
What Brought You to Andesite?
Before Andesite, I was working as a data science and analytics consultant at a small company in Austin. At the time, generative AI was blowing up, with everyone racing to implement it. My job required keeping pace with trends in data science, so I was getting to do some cutting-edge work. I gained a lot of experience deploying custom generative AI solutions. Andesite was exploring many of these same technologies. My experience matched the problems the team was trying to solve, so it was a great fit.
When I joined Andesite, much of the work was focused on open-ended technical exploration. We were thinking about how to apply emerging and existing technology in ways that were genuinely useful for our product. My background had made me comfortable working on problems that were not fully defined at the outset and figuring out how to frame them as something tractable. I enjoyed this unrestricted approach to problem solving.
At Andesite, one of the things I value most is that we’re trusted to explore ideas while still receiving support from our teammates. The culture here is highly collaborative. Our leadership team is so accomplished, but also approachable and humble. They value you, regardless of your title or tenure, and they notice the work you do, no matter how small.
How Did You End Up Working in Data Science? What Made You Choose This Field?
As an undergraduate at Johns Hopkins University, I studied cognitive science and English, which ended up being a surprisingly useful combination. One taught me to think about how complex systems like the brain work, and the other taught me how to analyze ideas deeply, examine them from multiple angles, and communicate my conclusions clearly.
What really changed my direction was research. I took a statistics class in college and, unexpectedly, I loved it. Around the same time, I started doing developmental psychology research and became fascinated by the process of asking an open-ended question and designing an experiment to answer it. It was like a fun puzzle that required creativity and clever thinking. That led me to pursue a master’s degree in quantitative methods at the University of Texas at Austin, where I learned how to apply statistical techniques to specific areas of research. While I was there, a professor suggested I look into data science, and it clicked immediately. It offered a lot of what I liked about research (experimentation, statistics, modeling, creating structure from ambiguity) applied to real-world problems.
What Are You Most Excited About With the Work You’re Doing Now?
I’m excited to build on our current platform to make it more useful without making it more intrusive. Our product uses AI to help cybersecurity analysts solve problems and investigate threats. I’m exploring how we can make the product feel like a natural extension of how these analysts already work. We want to help them move faster, catch things they might otherwise miss, and make better decisions without the AI getting in the way.
My focus at the moment is on growing our Evidentiary AI™ capabilities and improving our product’s memory to make it more attuned to customers’ environments over time. Our AI can remember how previous investigations unfolded and apply that context to a new situation. It’s part of the broader plan for our platform to have a rich understanding of an organization’s own data, behaviors, and history.
Another area I have worked on is dataset profiling. We automatically characterize a customer’s data using descriptive statistics and other summary information so both the analyst and the agent have a better understanding of what is available. That information becomes part of the customer’s data dictionary, where it can also be enriched with additional context. The goal is to offer a clear picture of the data the agent is reasoning over.
How Do You See AI Reshaping Our World, and How Will Data Science Reshape AI?
One of the most exciting things about AI is that it lowers the barrier between domain and analytical expertise. It’s meant to help us work smarter, not to take over our work.
As data scientists, part of our job is understanding the data well enough to ask the right questions. It can take a lot of time to build this expertise, especially if you don’t always work with the same kind of data. If I’m handed a dataset from a domain I don’t know well, I first have to understand what’s normal, what’s unusual, what the terminology means, and what questions are even worth asking. AI can accelerate that process dramatically. It helps me build enough context to get to the interesting questions faster, which is exciting.
Those who already understand a domain deeply often know exactly which problems matter, but they may not know how to design an analysis or write code. AI can help them bridge those technical skill gaps, allowing them to spend less time learning new tools or relying on someone with specialized expertise and more time investigating their questions.
While AI can help with the technical work, the hardest part is still deciding which questions are worth asking, whether the analysis actually makes sense, and what is missing. That kind of judgment comes from understanding the real-world context around the data, including the people, incentives, history, and constraints that may never appear in the raw data at all.
How Is Andesite Doing Things Differently?
One of Andesite’s strengths is how closely we stay connected to the problems our customers are trying to solve. We have teammates with deep experience across both the commercial and government sectors, which gives us a strong understanding of how security teams operate and where the real friction is.
What sets our product apart in the AI SOC space is the data analytics engine at its core, which is what our customers really need. That engine is the decision fabric our product is built on. It’s not just an LLM wrapper over a playbook. It can search across different data sources, reason over what it finds, and connect information that would otherwise be scattered across an organization’s data ecosystem. And it can show its work, so analysts can understand how it reached a conclusion and trace that conclusion back to the underlying evidence. The analyst is still driving the investigation, but Andesite does the heavy lifting.
It is especially exciting to me because I work with data in much the same way. A lot of the challenge is not simply retrieving information, but figuring out which data matters, how to connect it across sources, what context needs to be preserved, and how to present results without overwhelming either the user or the agent. Those are hard technical problems, and we have been thoughtful about balancing depth, usability, and transparency.
It all comes back to our philosophy of Humans at the Helm. We want the people using our product to see what information the agent is using, understand how it is reasoning, and challenge its conclusions when necessary. The more transparent that process is, the more useful, auditable, and trustworthy the system becomes.
Tell Us a Little Bit About Life Outside of Andesite. How Do You Spend Your Free Time?
Earlier this year, I married my husband, Robert, and we moved into our first home together. After wedding planning and house hunting all last year, I’ve been enjoying the quiet of settling into things together. I’m also a lifelong horse girl (fun fact: I had a horse named 27 growing up), and I still look for any excuse to be around horses. I enjoy Krav Maga and kickboxing, and when I want something more relaxed, I usually reach for a good book. Toni Morrison and J.M. Coetzee are two of my all-time favorite authors. Their work keeps me grounded in the beauty and depth of the human experience, which feels especially valuable at a time when so much attention is focused on AI.