How Virtu Financial is redefining AI data readiness

In our first episode of Data in the AI Era, Erin Stanton, AI and Data Leader at Virtu Financial, offers her insights in navigating the complexities of data readiness, strategy, and management in the AI era. Throughout the conversation, Erin reflects on her 15-year journey at Virtu, the transformation of their data strategy, and the deliberate steps they’ve taken to ensure their data fuels scalable and impactful AI initiatives.

Be first in line to hear our next podcast by subscribing to our YouTube channel, Apple Podcasts, Spotify or your chosen podcast platform. If you have any questions you would like us to tackle in a future episode or want to provide some feedback, reach out to the team at [email protected].

Show notes

The conversation begins with Erin recounting her early days in finance, a time before machine learning was a buzzword . She explains how her team’s passion for thoughtful decision-making—asking not just “can we deploy AI?” but “should we?”—has guided their success. Erin walks us through Virtu’s pivotal decision in 2016 to rebuild its analytics framework, laying a foundation that fortuitously positioned them for AI innovation years later.

“AI-ready data isn’t just about quality,” Erin explains. “It’s about curating it for use cases we can’t yet imagine.” This perspective reflects the growing demand for organizations to rethink their data strategies, ensuring they can fuel reliable, scalable AI initiatives.

As the discussion unfolds, Erin shares how Virtu ensures data is AI-ready through transparency, rigorous preparation, and a willingness to learn from failure. From hosting monthly ‘Machine Learning Mondays’ to manually curating datasets, Virtu has nurtured a culture of collaboration and precision. Erin’s insights highlight how the right data strategy can empower organizations to harness AI’s potential without losing sight of business value or risk considerations.

If you’re curious about how to make data work harder, smarter, and more transparently for AI, this episode is a must-listen. Watch the full episode below or read through our key takeaways to help you on your own journey to AI success.

Quotes to remember:

“AI-ready data isn’t just about quality—it’s about curating it for use cases we can’t yet imagine.”
“Good data, good processes, and good judgment—that’s what drives AI success.”
“Transparency fosters trust and dialogue, opening the door to better solutions.”
“There’s no shortcut to truly understanding and preparing your data.”

Key takeaways:

1. Build a strong foundation for AI

Early strategic choices can significantly accelerate your AI scalability.

Virtu’s 2016 decision to overhaul its analytics framework set the stage for its AI evolution. By integrating time-series data tools, Virtu built a centralized, API-enabled data ecosystem that now powers AI applications across pre-trade, real-time, and post-trade workflows.

“We got lucky with our early technology choices,” Erin admits, “but those choices supercharged our ability to scale AI today.”

2. Prioritize contextual data preparation

Prepare data for AI by curating datasets that reflect the specific nuances of your business challenges. Go beyond cleaning and labeling—focus on feature engineering and advanced filtering to align data with real-world problems. Erin recounts a three-year journey to improve model accuracy by refining training data through feature engineering and advanced filtering techniques.

“AI-ready data isn’t just about accuracy; it’s about reflecting the reality of the problem you’re solving.”

3. AI readiness requires sweat equity

Don’t underestimate the manual effort required for robust AI. For their latest GenAI project, Erin’s team spent months manually tagging data to ensure contextual accuracy. “Getting data AI-ready isn’t just about automation—it often requires human effort to make it robust.”

4. Embrace transparency to build trust

Implement regular review initiatives like open forums or collaborative sessions to share and discuss models.

Virtu Financial fosters trust through initiatives like “Machine Learning Mondays,” where clients and teams openly review models and algorithms, including those that failed. Erin emphasizes that transparency is not just about showcasing success but inviting feedback. “We’ve had clients point out things we missed, and that’s been invaluable.”

5. Foster collaboration and continuous learing

From Coursera-based learning programs to automation hackathons, Virtu invests in upskilling teams and fostering cross-functional collaboration. “Good data management isn’t siloed—it’s a team sport,” Erin asserts.

6. Adopt a riskaware approach to AI

Evaluate AI initiatives with a focus on mitigating reputational and regulatory risks. Ensure data and AI models are robust enough to make trustworthy decisions, as there’s no substitute for accuracy and reliability in high-stakes environments.. Erin stresses the importance of a risk-aware approach: “Your data must be robust enough to ensure AI decisions are trustworthy—there’s no shortcut here.”

7. Know when to say no to AI

Assess AI opportunities critically. Avoid deploying AI for the sake of it; instead, focus on areas where it adds meaningful value. Be transparent about decisions not to use AI, as this can strengthen credibility and strategic focus.

In a world where AI is often overhyped, Erin champions the value of restraint. “Sometimes the best decision is not to deploy AI. Knowing when to say no is just as critical as knowing when to innovate.”

“Being public about why and where we don’t use AI is just as critical as showcasing its successes,” Erin adds.