From using LLMs to judge other LLMs to harnessing unstructured data, from building better recommendation systems to anxiety management—Data Science Salon Austin came with groundbreaking speeches, inspiring panels, and a startup showcase.
The Data Science Salon Austin conference focused on building, applying, and implementing AI solutions in the enterprise. To further reduce barriers between practitioners and aspiring disrupters, the conference featured a Startup Showcase, where companies with innovative products were invited to show how their solutions will change the world and the existing order.
The Festival of Knowledge Sharing
The first day of the event was full of talks and discussion panels. The conference opened with a speech from Anna Anisin, CEO and founder of Data Science Salon, who welcomed the guests.
Building Better Healthcare
Aarohi Tripathi, Senior Data Engineer from CVS Health, shared insights on how to use AI-powered innovations in healthcare. She discussed how AI may improve diagnostic efficiency and accuracy, providing better patient care and more narrow diagnostics. “Everyone is different. That’s why we need personalized healthcare. Not a generic approach, but care that is tailored to this one, particular person:” she said. She also covered how to implement and utilize AI solutions to speed up medical research and reduce costs.
Her speech was followed by David Talby, CEO at John Snow Labs, who presented his perspective on integrating Multi-Modal, Reasoning, and Conversational Medical LLMs to Understand Complex Patient Stories. Focusing on Large Language Models, he showed how these new tools improve quality and security in healthcare applications.
Continuing the topic of AI solutions in healthcare, Ari Heljakka, founder and CEO of Root Signals, explored managing AI agents and succeeding with Strong, Swift, and Affordable LLM Judges. One key takeaway was the need to build reliable, trustworthy AI applications at scale.
Enhancing Enterprise Operations
After the coffee break, Dhivya Nagasubramanian from US Bank provided insights on Harnessing the Potential of Unstructured Data to Drive Enterprise Success. According to her, unstructured data accounts for about 80% of all enterprise data—emails, text, images, and similar information. Using AI-powered solutions to analyze these data types is crucial for making them actionable and business-relevant.
Her talk was followed by Arthur Keen, Senior Solutions Architect at ArangoDB, who shared his remarks on Unlocking the Power of Graph-Based Retrieval and Analytics for Unstructured Data.
Next came a panel on Generative AI: Seizing Opportunities and Overcoming Challenges in the Enterprise. Experts from AMD, Comcast, Viasat, and others discussed the possible applications and challenges for AI in enterprise settings. They also emphasized how unstructured data remains a challenge for current solutions and can be addressed through dedicated database systems.
Seeing the Future—and Making It Better
After lunch, Dr. Greg Michaelson—Chief Product Officer & Co-Founder at Zerve—shared how data science shortened the COVID-19 pandemic and reduced the number of casualties. He outlined data-driven strategies used to tackle the pandemic and better allocate resources.
Arun Kuppam, Director of Data Science at Gap, Inc., spoke about Revolutionizing Forecasting: A Comparative Analysis of Time Series Methods Across Industries, offering his perspective on the impact of LLMs in the fashion industry—and beyond. He highlighted how AI can support logistics, sales, and various day-to-day operations in retail, elaborating on less common tools like hierarchical forecasting. “Hierarchical forecasting can also be used in retail, for example in e-commerce, to forecast a traffic website, and separate categories. Usually these forecasts don’t stack, and that’s where hierarchical forecasting is applicable. It is about having a forecast on every level” he says.
Fatma Tarlaci, chief AI officer at SOAR AI, was the next speaker, discussing Agentic AI for Data Science and expanding on the directions initially covered by earlier speakers. “But do we actually need agents?” she asked during her speech. “Prompts turn out to be not so great. If we use all the business knowledge we have, we end in “prompt and prey" scenario. And Agents decompose problems into logical steps” she answers right away.
Kailash Thiyagarajan, Senior Machine Learning Engineer at Apple, provided insights on Enhancing Recommendation Systems With LLM-Generated Aspect Augmentations. He explored how modern LLMs address a primary challenge of traditional recommendation engines—lack of context, which reduces their effectiveness.
Arjun Bali, Senior Data Scientist at Rocket Mortgage, spoke on Advanced RAG for Text-to-SQL Applications, focusing on real-life RAG implementations for handling complex financial data.
The Startup Showcase
On the second day, in addition to more talks, a startup showcase took place. Panels and discussions revolved around building and scaling AI-powered startups in the enterprise environment.
The day kicked off with a Fireside Chat featuring Christine Galib (Director, Advancing Photonics Technologies at Princeton University), Garnet S. Haraman (Co-Founder, Managing Partner at Aperture Venture Capital), and Rohin Tagra (Founder & CEO at Azimuth). The experts discussed How AI Startups Can Meet Evolving Investor Expectations.
The chat was followed by a panel discussion with Christine Galib (Director, Advancing Photonics Technologies at Princeton University), George Ploss (Head of VC Practice at Oracle), Gopinath Sundaramurthy, Ph.D. (Founder & Head of Data Science at Ensemble VC), Oksana Malysheva (Managing Partner & CEO at Sputnik VC), and Jesse Martinez (Founder & CEO, Chair at LSA Global HQ). The panel focused on Capitalizing on Creativity: Venture Capital in the GenAI Boom.
The second day also marked the conclusion of the startup showcase. ScalafAI was the winner of the contest, while Jevelwision took runner-up honors.
Summary
At Data Science Salon Austin 2025, hundreds of participants and speakers delved into the cutting edge of machine learning and generative AI. Acting as a pivotal guide to AI’s rapidly evolving landscape, this event was a rare opportunity to exchange ideas in a non-discriminatory, open, and inclusive environment. It also provided valuable chances for data science practitioners to forge new connections. The conference was held at Oracle’s headquarters in Austin.