In this interview, we sit down with Anirudh Reddy Pathe, Senior Director of Decision Science at Glassdoor, to explore how data-driven decision-making is reshaping business strategy. From his journey spanning engineering, fintech, and travel to leading decision science at Glassdoor, Anirudh shares practical insights on building high-performing teams, aligning AI initiatives with business outcomes, and fostering a culture of experimentation. He also reflects on lessons learned from failure, signals of organizational readiness for AI, and how the role of decision science may evolve with the advent of generative AI.
Explore more interviews here: Humans in the Loop: Five Voices Redefining the Future of AI
Your journey from engineering to leading Decision Science at Glassdoor is quite dynamic. Can you walk us through the pivotal decisions or inflection points that shaped this trajectory?
My journey has really been about following curiosity and leaning into moments where I could have more impact. Early in my career, I was deeply technical, focused on engineering and problem-solving at the system level. The inflection points came when I realized effective utilization of data can solve significant business problems at scale. At Priceline, I saw the power of rapid experimentation. At Discover, I saw how large teams can move rapidly together if a strong tech team and a strong business team combine forces. At Glassdoor, I deliberately pivoted toward leadership in Decision Science because it sat at the intersection of technical depth and business decision-making.
You’ve built and scaled global teams across companies like Discover and Priceline. What are your core principles for building high-performing data science teams in today’s remote and hybrid environments?
I anchor on three principles: clarity, autonomy, and connection. Clarity around mission and success metrics so teams know the “why.” Autonomy so talented people have room to solve problems their way. And connection, especially in hybrid setups, so no one feels like a lone satellite. I invest a lot in rituals that make distributed teams feel cohesive, whether that’s shared OKRs, cross-team project ownership, or simply creating space for informal interaction.
How do you ensure that AI and automation initiatives stay closely connected to real business impact and not just become ‘shiny objects’ within organizations?
The question I always ask is: What outcome will this change? If an AI model doesn’t clearly tie to a lever the business already cares about, revenue growth, cost efficiency, and user experience, it risks being shelfware. I push teams to define success in business terms first, then scope the AI solution. That discipline keeps us from chasing shiny objects
With decision science touching so many functions, product, marketing, operations, how do you prioritize and balance cross-functional needs without diluting focus?
Decision Science sits in the middle of everything: product, marketing, ops—and it’s easy to get stretched thin. I’ve learned to prioritize where the flywheel effect is strongest. If an insight or experiment unlocks compounding value across functions, that’s where I’ll invest. I also use transparent roadmaps so stakeholders see trade-offs, which helps manage expectations
What do you see as the biggest misconceptions leaders have about experimentation, and how do you coach them through adopting a test-and-learn culture?
One misconception I run into a lot is the belief that every experiment has to deliver a “win” to be valuable. In reality, most tests don’t move the needle in the way people expect, but that doesn’t make them failures. Each one sharpens our understanding of customer behavior and guides us toward better bets. Another misconception is that experiments are only about UX tweaks or surface-level optimizations. The truth is, when done right, experimentation can influence strategic choices, what markets to enter, what products to sunset, and how to allocate investment. Helping leaders see experimentation as a decision-making framework, not just a product tool, changes the conversation entirely.
Your work has spanned different industries from fintech to travel to tech. How does your approach to data-driven growth adapt across such varied business contexts?
Whether it was fintech, travel, or tech, the constant has been using data to reduce uncertainty and create value. The adaptation comes in what risk and opportunity look like in each domain. In fintech, risk was litigation and governance, but the opportunity was immense to drive financial outcomes for millions. In travel, it was a conversion under intense competition, and the opportunity was to provide memorable life experiences for millions of people. At Glassdoor, the risk is user trust, but the opportunity is the ability to help millions of people find the job they love. The playbook is the same: clarify the decision at stake, then bring the right blend of analytics, experimentation, and storytelling to influence it.
How do you see the role of Decision Science evolving over the next 5 years, especially in the wake of rapid advances in generative AI?
I think Decision Science will evolve into being the connective tissue between AI and the business. With generative AI lowering the cost of analysis, the differentiator will be who can ask the right sets of questions and translate them into action. Five years from now, I expect Decision Science leaders to be less about building models/ analysis themselves and more about orchestrating humans and machines to drive business outcomes.
Can you describe a time when a data-driven insight conflicted with executive intuition, and how you navigated that tension to drive alignment?
I’ve definitely been in rooms where the data contradicted gut instinct. One example: we tested a feature executives loved, but the experiment showed it actually reduced engagement. Navigating that tension required empathy, acknowledging the intuition while being clear on the evidence. In that case, showing the longer-term impact through simulation helped build alignment. The key is framing data as a decision partner, not a decision dictator
What signals do you look for when assessing whether an organization is truly ready to embrace AI-enabled decision-making at scale?
I look for three signals: leadership alignment on the why of AI, a culture that values testing over certainty, and data foundations that are trustworthy. If any of those are missing, scaling AI becomes a struggle. You can hack your way into a proof-of-concept, but sustained impact needs those ingredients
If you were designing a graduate-level curriculum for Decision Science leaders of the future, what three courses would be mandatory and why?
Great Question! I’d design it around:
Causal Inference & Experimentation – the backbone of good decision science.
Organizational Behavior & Influence – because insights don’t matter if you can’t drive adoption.
AI Ethics & Governance – leaders need to understand not just what AI can do, but what it should do.
Tell us about a failure or false start in your career that ultimately unlocked growth or clarity.
Early in my career, I pushed for a massive model overhaul at Priceline without fully aligning stakeholders. Technically, it was elegant; organizational, yet it was dead on arrival. That failure taught me that influence and alignment matter as much as technical correctness. It changed how I lead ever since
- What’s a book, framework, or piece of advice that fundamentally reshaped how you lead teams or drive change?
- If you had to create a decision-making “dashboard” for your own career, what would be your top three metrics or signals to monitor?
If I built one for myself, the three metrics would be:
- Impact – am I moving the needle for the business and the people I lead?
- Learning velocity – am I stretching myself in new ways?
- Growth – am I able to grow people on my team
