Image
Why Data Needs Context: Lessons from Anaya Nath’s Internship at Sotheby’s
What happens when a lifelong love of art meets an aptitude for math and data science? For Anaya Nath, the answer led to a summer internship at Sotheby’s.
A double major in data science and economics, and a member of the School of Data Science’s inaugural B.S. in Data Science class, Nath spent the summer working with Sotheby’s Data Engineering and Business Intelligence team, where she helped build AI-powered tools designed to turn complex data into useful insights.
The experience put lessons from the classroom into practice. From data engineering and infrastructure to communicating across technical and business teams, Nath saw firsthand how data science and AI can create value in an industry steeped in history, culture, and creativity. And for Nath, it reinforced something she had begun to discover at UVA: the most interesting problems often live at the intersection of disciplines.
Q: Tell us a little about yourself. What drew you to UVA and the B.S. in Data Science program?
When I was applying to colleges, UVA stood out to me for its spirit of innovation, intellectual rigor, and strong sense of community. I read about Professor Ken Ono, who was using mathematics to measure and track swimmers' performance. I grew up swimming in Mumbai, India, where I'm from, and I loved the analytical rigor of math. While swimming is an optimization problem in itself, I'd never really thought about it that way until I came across his work. That connection really stayed with me.
I knew I wanted to major in some flavor of math, drawn to how it works from first principles: proving formulae through logical assumptions and reasoning through solutions. But I wanted to keep an open mind my first year and sit in on classes I knew nothing about.
So that fall, I attended DS 1001: Foundations of Data Science and really enjoyed the format: understanding data science through systems, design, value, and analytics. Guest speakers translated this theory into real-world applications: Neal Magee introduced us to cloud infrastructure, Renée Cummings to machine bias and discriminatory models, and Cathy O'Neil to the algorithmic harms she writes about in her book "Weapons of Math Destruction." It made me see data through a holistic, interdisciplinary lens, which really sparked my curiosity about how technology and society shape each other and what that means for progress.
Q: You're part of the inaugural BSDS graduating class. What has it been like to be part of the first bachelor's cohort?
Being part of the inaugural BSDS class, at a moment when the world and the industry are still figuring out what data science will morph into as AI advances so quickly, is really exciting, but also a little scary. The School has been great about staying open to conversation and change: They genuinely want to know what is and isn't working for us as students and how they can do better.
My cohort is awesome. We see each other almost every day, and through shared projects, lectures, and even random conversations, I've learned a lot about what my peers are doing, which has helped shape my decisions and strategies.
My biggest takeaway has been how accessible opportunities are: research conferences, funding, access to brilliant professors, and the freedom to pursue concentrations that genuinely interest me. That's given me a breadth of exposure across various aspects of the data science field that I'll carry with me when I graduate and enter the industry.
Q: Tell us about your internship with Sotheby’s. What attracted you to the opportunity, and what does your role involve?
The best advice I got for my job search was to be creative and really look for alignment. I found out that Sotheby's, the art auction house, had a technology internship through a chance conversation with a peer and decided to explore it. It seemed like a natural fit. The art market as the context, and data and applied AI as the mechanism: building tools and optimizing the processes Sotheby's uses to create value within it.
Art has been a throughline for me since I was young. I had solid training growing up, working with different teachers across a whole range of mediums to build my own portfolio: pencil sketches, watercolor, acrylic, oil, prints, graphics, sculpture, pen-and-ink. Moving between them taught me to notice the small details: the weight of a line, the layering of a wash, the quiet decisions that give a piece its nuance, and that's what gave me the ability to really appreciate art and everything it holds.
My role at Sotheby's was in the Data Engineering and Business Intelligence department on the Product and Technology team. There's so much rich data at Sotheby's about buyers and sellers, the artists, and the pieces themselves, considering their provenance, medium, condition, value, and estimates.
Zooming out from my own experience making fine art to understanding the art market and Sotheby's place in it, through the lens of that data, research, and economics opened up a whole new world for me.
Q: What kinds of projects or problems have you been working on this summer, and what has been the most exciting or rewarding part of the experience?
Sotheby's is in the middle of a major transition from BigQuery to Snowflake as its main data warehouse, so my first project was building an AI agent in Snowflake: a conversational layer over a dataset that previously relied on a rigid Tableau keyword search. It is meant to be used by members of the client strategy team. The chatbot has three parts: a Cortex search service that points to the source table and returns a key to the semantic layer, which contextualizes that key within the vocabulary it's been taught, all orchestrated by an agent.
Much of the work was talking with data engineering and business intelligence members to understand their priorities and make sure the chatbot was as useful as possible for the end user while optimizing for cost and existing compute resources.
What made this especially exciting is that Sotheby's is part of Snowflake's pilot program, so I got to work directly with a Snowflake representative to map their existing AI capabilities onto my problem and find creative workarounds to meet many conflicting stakeholder needs.
My second project was a Streamlit app that lets the logistics team quickly look up key inventory details for operations planning. We also had a capstone project at the end where interns from different departments ranging from logistics and evening sales to contemporary and Asian art worked together to identify a new business or audience Sotheby's could tap into.
On my own, I could pull the numbers and see the patterns, but they only really came to life once we thought through them together: Someone's read on the evening sales calendar or the contemporary market would give a figure a meaning I'd never have reached alone. It was a great way to see how data only becomes insight when it's grounded in the relevant context.
Q: How has the BSDS program prepared you for this internship?
First, the ability to understand diverse data infrastructure. Professor Neal Magee's Data Engineering course (DS 3022) gave me a solid grounding in working with different data formats, batch processing, automated pipelines that execute a series of coordinated steps, and stream ingestion for real-time analysis. These concepts helped me understand the diverse data sources that Sotheby's infrastructure draws from, how a change to one table can have upstream effects, and how to engineer a consolidated, enriched dataset given that data arrives from different sources at differing intervals.
Second, communication: being a liaison between technical and non-technical stakeholders. In my internship, I worked closely with the data engineering lead and the business intelligence strategist, who approached the same tasks from two different perspectives. Communicating with a business stakeholder in a way that's clear yet simple came fairly naturally to me, since in Professor Prince Afriye’s Design I: Communicating with Data class, we explained our projects to classmates who had no prior context. His whole premise is to explain what you learn: the models, the algorithms, the underlying math, as simply and intuitively as possible, and that has really stayed with me.
Third, the ability to see databases as dynamic, interconnected entities. Professor Peter Alonzi's Data by Design (DS 4320), where we built entity-relationship diagrams and learned the importance of hard and soft metadata, documentation, and schema, was key. It directly helped me hand off my data assets to the rest of the team when my internship ended.
Q: What technical skills have you relied on most, and what professional skills have been just as important to your success?
Technically, I relied most on Python and SQL: writing queries, building ETL data pipelines, and making deliberate choices about data visualization, always weighing accessibility and interpretability so the output actually lands with the person using it.
On the professional side, communication mattered just as much, but so did managing a project across its full lifecycle: clarifying the initial brief so I was solving the right problem, keeping my manager updated as things evolved, and making sure the final output met the specification. Staying in close communication at every stage, not just at the start and end, kept the work on track.
Presenting was a big part of this too: I learned to think of it as building a story by stitching together the data, the context, and the decisions into a complete picture, so the audience doesn’t just see results but understands how everything comes together.
Q: How have you been able to apply what you've learned in the BSDS program to your work at Sotheby’s?
One moment where the classroom and the work really converged was when I had to decide how to structure the data behind my AI agent. In DS 3022, I learned that data lands in a warehouse at different times and that a change to one table can ripple upstream into everything built on top of it. At Sotheby's, my chatbot needed to pull from several sources that updated on different schedules, and a naive setup would have been fragile and expensive to query.
Working with the data engineering lead, I used that framing to reason through the options, and we decided to build a single, enriched dataset, one that folds in those updates as they arrive, for Cortex to point directly to. This helped put into practice the theory we learned in class. It was the moment an abstract concept from class became the actual thing standing between a brittle prototype and something the client strategy team could rely on.
Q: You received SELF funding from the School of Data Science. How did that support enhance your internship experience or professional growth?
I received SELF funding for an unpaid internship in the spring of my third year at Table22, a technology startup building a restaurant e-commerce platform that helps restaurants and food vendors monetize their products through subscription-based models like wine clubs and meal kits. Because the role was unpaid, the funding was what made it possible for me to take it on at all, and it also helped cover relocating to New York for my next internship at Sotheby’s.
The internship was a great early look at how a startup turns data into business decisions. My work centered on which wine club offerings drove the most value, measured through operational and financial metrics like AGMV and conversion, which taught me to connect analysis directly to the levers a business actually cares about.
Q: Have experiences outside your coursework shaped your growth?
I'm on the executive board of the Business Ethics Society: I was director of scholastic development last year and I'm vice president this year, and it's really opened me up. It's given me the confidence to share my perspective and argue for what I believe in while still looking for merit in the counterargument. It's taught me to see situations through a lens of ethics, considering purpose beyond profit and thinking about individual, societal, and global impact.
I've also attended a lot of tech conferences through the School of Data Science, like the Applied Machine Learning Conference and the All Tech Is Human Conference in NYC, which I first heard about through Professor Mona Sloane. Hearing leaders in the field share their work is fascinating because it illustrates what progress could look like, where the frontier is heading, and how industry, research, and society intersect. It's forward-looking and deeply rooted in scientific rigor and deep expertise, and that's what I find really compelling.
In grade 12, I compiled my portfolio of illustrations into a self-published activity book, stART! Synaptic Transformations Through Art. I wanted to make art accessible to those who considered themselves 'not good at it,” so I delved into research from the fields of neuroscience, psychology, and art therapy and created research-backed art activities to encourage young adults to use guided art as a tool for mindfulness, calm, cognitive agility, and focus.
I'm also part of the UVA Real Estate Professional Network through the McIntire School of Commerce, where I've met some really interesting mentors who've offered guidance and encouraged me to learn about PropTech, which has become an industry context I'm curious to explore.
Q: As you prepare for your fourth year, how has your summer internship experience influenced what’s next?
My internship gave me a lot of clarity that I want to commit to a career in technology. I went into Sotheby's expecting to do traditional data engineering and pipeline construction, but working with AI agents and Streamlit apps inside the Snowflake environment showed me that a consolidated data infrastructure is the best foundation for automation. So I now know I want to work in data science and data engineering roles that leverage agent capabilities.
When I think about it, all my interests tie together nicely. In high school I loved literature and have always been fascinated by rich vocabulary and semantics, and I also love math. Working with AI agents sits right at that intersection: teaching a system to understand language and meaning, grounded in the logic and structure of math. I'm excited to see what's ahead.
Q: If a prospective student asked why they should consider the BSDS program, what would you tell them?
If you like thinking at the intersections of disciplines and approaching problems holistically, BSDS is for you. It's a program that trains both sides of your brain: the creative and the logical, and that balance gives you a genuine overview of the entire data pipeline, along with a foundation for understanding where technology and AI are heading. I think it prepares you on two levels: not just as a student entering the job market, but as a citizen navigating a world being reshaped by technology.
Q: Finish this sentence: “The BSDS program prepared me for my internship by…”
Teaching me to work across both the technical and the human sides of a problem: to build a pipeline and explain why it matters.
Discover the data science bachelor's degree at the University of Virginia. Request more information, connect with Admissions, or start your application today.
Learn more about the Stephenson Bicentennial Scholarship Fund here. The Stephenson Experiential Learning Fund (SELF) supports high-impact, hands-on experiences and opportunities for BSDS students including Career Treks, Internships, Undergrad Research and Micro-Internships. For more information, please contact the Office of Student Affairs Office of Student Affairs at stephensonscholars@virginia.edu.
