Office Hours: I’m the Head of Research at Tableau. I’m Vidya Setlur. AMA!Featured

Posted in:Software Engineering Office Hours Management and Leadership

Hi Elphas!

I’m Vidya Setlur, the Senior Director of Tableau Research. My research interests lie at the intersection of natural language processing (NLP) and computer graphics to better understand data semantics and user intent to inform the meaningful visual depiction of data.

Since joining Tableau Research in 2012, I have worked on various projects and prototypes involving semantics and visual representation. I have also been deeply interested in exploring analytical conversation using natural language interaction via a system I developed called Eviza.

Subsequently, as part of productizing Tableau's natural language feature, Ask Data, I moved to engineering and was a manager on the Natural Language Team at Tableau. I recently co-wrote a book, "Functional Aesthetics for Data Visualization."

Before Tableau, I was a principal research scientist at Nokia Research Center for seven years, where I worked on intelligent retargeting of photos on small mobile displays and map navigation systems.

Prior to that, I did my PhD at Northwestern University, graduating in 2005. I spent the formative years of my life growing up in Boston, New Orleans, and Arizona, and then moved to Bangalore, India, to finish up high school and undergrad.

During my downtime, I enjoy gardening and hiking, as well as performing classical Indian dance.

I love good coffee and trying out new baking recipes.

Ask me anything about my research and career journey - being an industry research scientist, how NLP and AI can help people see and understand data, the process of writing a technical book, or being a woman in tech. Looking forward to hearing from you!

ElphaStaff·Jul 29

Thanks so much for joining us @vidyasetlur!Elphas – please ask @vidyasetlur your questions before Friday, August 2nd. @vidyasetlur may not have time to answer every questions, so emoji upvote your favorites 🔥👍🏾➕

Iynna·Aug 01

Thank you for being with us this week, Vidya! I am curious what prompted you to do a PhD initially? Were you thinking of going into research and how did you come across your first work opportunity?

vidyasetlur·Aug 03

Hi! I've always been intellectually curious and enjoyed coloring outside the lines, so to speak. For me, pursuing a PhD was an opportunity to explore my curiosity, combined with my love for writing to articulate my work and share with the larger community. My first work opportunity at Nokia Research happened by accident. My husband put my resume on ACM to get a sense of the job market, and Nokia reached out to me as my dissertation concerned retargeting large-resolution imagery to small displays such as mobile screens. I decided to interview, got the job offer, and my advisor was understanding enough to let me graduate a year early.

MichelleFuentes·Aug 01

Thanks so much for taking the time to answer elphas' questions, Vidya! It's so cool to hear your journey and the things you've accomplished! I'm curious on if there are any specific things you've worked on in the past few years that you never would have guessed you'd get to work on back when you first joined Tableau?? (It must be cool to do a look back at everything you've developed and worked on since you first started - like "Functional Aesthetics for Data Visualization", Eviza, and more)

vidyasetlur·Aug 03

Hi Michelle, when I first joined Tableau in 2012, I never would have guessed that NLP would become a household term in the analytics space. Now, fast forward a decade, we are exploring ways in which generative AI can be used to further the mission of helping people see and understand data. Looking forward to the next decade of innovation in this space!

Kalyn179·Aug 01

hi Vidya! thanks for your time. what's the process of writing a technical book like? in what ways does it differ from other types of books? excited to hear your insights! thanks!

melaniealmeida·Aug 06

Thank you for this question. I had the same one!

vidyasetlur·Aug 03

Writing a technical book involves long hours of outlining, researching, and writing (with copious amounts of coffee :)) . The process involves distilling complicated concepts in a way that is understandable to the target audience, along with relatable examples and clear figures. There is also a lot of work put into citing related work and making sure the content is accurate and frames fresh ideas and perspectives that differentiate the book from others on a similar topic. And always helps having an editorial team to review the manuscript.

melaniealmeida·Aug 06

With retrospective, do you think this work investment was worth it?Did your company arrange your schedule to help you fit writing time?

bridgetcogley·Jul 29

My favorite co-author! 💖Eviza was truly groundbreaking work. GenAI and NLP are radically transforming how we all are thinking about product roadmaps. Beyond what we see today, where do you see GenAI going? What else do you predict? What's something you've researched from our book where the findings surprised you? Lastly, what's your favorite ZombieRunner roast?

melaniealmeida·Aug 06

Hello @bridgetcogley, can you elaborate why you think GenAI and NLP will transform Product roadmaps? As a Product Manager, I feel concerned about your point, but I haven't investigated yet much on GenAI and NLP. I am not an early adopter type of person.

bridgetcogley·Aug 06

I play in a space that's more on the early adoption side. Specifically around Gen AI and NLP, I'm seeing work (Vidya's research is a great example - definitely take a look: https://www.tableau.com/research/people/vidya-setlur) around refining the search experience as a big one - yes, providing sometimes and answers, but also moving past the bag of words ethos ("stores near me shoes running" and then using filters for prices) to shifting to dialectical ("where can I find the best running shoes at a mid-range price point?"). Forms are another area - where do people currently write paragraphs that go to die in the CRM? How can GenAI start to better structure that data? I watched someone today type an entire med list into open text - that could be parsed and structured in a way that's easier to use downstream without forcing the person doing the entry work into another rigid form format. That's some of the smaller cases I'm seeing. Larger ones are some of the call summarization tools - listening to calls, transcribing them, and then doing various labeling and analytical tasks on the discourse itself. How effective was the sales pitch? Where was this push back on this call and against the broader trends? Is this a product problem? AI is filling this space. I'll end on this video, which really landed with me. Arvind talks about intelligence as agency and this may truthfully provide the best answer to your question. https://m.youtube.com/watch?v=Do3opAtF4W4

vidyasetlur·Jul 31

Bridget! Yes, my partner in crime in writing the book together and several follow-up papers. Great questions. You are right about GenAI being a radical game changer in how we think about our tools and workflows. For starters, I foresee text being elevated as a first-class citizen in visual analytics, thanks to interfaces and enablers that use genAI to provide narrative data story outlines, summaries, annotations, alt-text, and more. Tasks that are often tedious, such as data cleaning, will be automated, enabling humans to do more of what they thrive in doing, i.e., being creative, curious, and collaborative. I predict new workflows emerging as humans continue to innovate, with data insights meeting where people are.Regarding the book, I remember us having a small chapter on text with charts, but boy, that topic has provided a rich trajectory of some really cool follow-up work where we found that users prefer lots of text with charts, challenging earlier premises in our community about using text sparingly and keeping the ink-to-visual ratio, low. I see more innovation emerging in this space that will further bridge the researcher and practitioner communities. And yes, Indian Mysore coffee beans with a medium roast, is my favorite cup of joe.

divyahegde·Jul 30

Hello Vidya! I'm trying to break into a Data Science/Machine Learning engineer role. What would be your tips for me to do so? How is a research career different from a traditional career path? And what does it take to be in a research based job?

vidyasetlur·Jul 30

Hi Divya, thanks for stopping by! Getting into a data science or NL engineering role requires building a strong foundation in relevant education and technical skills. Research careers often differ from traditional paths by focusing on innovation through running data science studies, publishing findings, and working with academic collaborators. Traditional careers at a company often focus on the application of your skills to solve customer problems, influence product roadmaps, and help build and enhance product features. Success in research requires relentless intellectual curiosity, strong analytical skills, and a good publication record.

RoDixon·Jul 29

Hi Vidya, you have such an interesting role/career journey! Thanks for sharing and answering questions. I was wondering if you had any recommendations for offering mentorship? I’m a lead UX Designer with 15 years of experience as both an in-house designer and consultant- and finally feel worthy of the term “expert” and would just like to share that and help the next generation of UXers grow. I thought having the combination of PhD & leadership role you might have some advice in this area. Best of luck! ☘️

vidyasetlur·Jul 31

Thank you for your kind words! It's fantastic that you're looking to offer mentorship and pay forward. To start with, I think it is important to understand your mentees' backgrounds, their goals, and challenges, and create a safe space for open communication. For some mentees, it's navigating career progression, while for others it could be handling a challenging work situation, or trying to balance work and life. Help them set clear, achievable goals, and encourage them to set the agenda when you meet with them. Share your relevant experiences, including successes and failures (this one is key!), and provide specific, actionable feedback while balancing critique with positive reinforcement. Finally, regular check-ins are essential to maintain an effective mentorship relationship.

share this article