“I still remember the smell of coal in the air,” Saurabh Kumar says, describing his childhood in Bokaro, a steel township in Eastern India. “Growing up in an industrial town, you understand the power of data before you even know what to call it. The steel plant engineers would predict production output based on hundreds of variables—that was my first glimpse of what would become my life’s work.”
That early exposure to industrial-scale data analysis would prove invaluable in Kumar’s groundbreaking work at NASA, where he tackled one of the most pressing environmental challenges of our time: predicting and preventing large-scale fires across continents.
A Global Challenge
At NASA’s Earth Science Division, Kumar led a pioneering project analyzing the Global Fire Emission Data datasets spanning 1997-2015. The scope was ambitious: studying fire patterns across Sub-Saharan Africa, Boreal and Central Asia, South and South-Eastern Asia, Australia, South America, Canada, the USA, and the Delhi-NCR region.
“The complexity of the data was overwhelming at first,” Kumar admits. “We weren’t just looking at one region or one type of fire pattern. We were analyzing burning patterns across vastly different ecosystems—from the Boreal forests of Canada to the bushlands of Australia.”
His team developed sophisticated analytical models that could process and interpret annual and monthly average burned areas across these diverse regions. “What made this project unique was our approach to Principal Component Analysis (PCA),” Kumar explains, his eyes lighting up. “By analyzing the spatial and temporal patterns of burned areas, we could identify critical predictive factors that varied significantly by region.”
The project’s impact was immediate and far-reaching. The seasonal and monthly burned area maps they created became crucial tools for fire prevention agencies worldwide. “When we started seeing how our predictions were helping firefighters in South America position their resources more effectively, or how Australian fire services were using our models to prepare for bushfire seasons—that’s when I truly understood the power of data science to save lives.”
From Environmental Protection to Financial Inclusion
Kumar’s success at NASA caught the attention of Capital One, where he would apply his expertise in pattern recognition to an entirely different challenge: democratizing access to financial services.
“The principles were surprisingly similar,” Kumar notes. Just as we looked for subtle indicators of fire risk in environmental data, we began identifying non-traditional indicators of creditworthiness.” His team pioneered the use of alternative data sources in credit scoring, enabling Capital One to provide financial products to millions of previously underserved individuals.
“We incorporated variables like utility bill payment history and rent records—data points that told a more complete story about financial responsibility,” he explains. The initiative generated over $100 million in revenue while giving 45 million individuals access to better financial products. “That’s when I realized that the pattern recognition skills I developed tracking forest fires could literally change lives in completely different contexts.”
Revolutionizing Urban Mobility at Lyft
At Lyft, as Head of Data Science for TBS Payments, Kumar faced a new challenge: enhancing payment security while improving accessibility in the company’s bike-share program. Drawing on his experience with global data patterns, he developed fraud detection systems that could distinguish between genuine financial constraints and fraudulent activity.
“Working with NASA’s global datasets taught me to look for regional patterns and cultural contexts,” Kumar explains. At Lyft, this meant understanding why payment patterns varied so dramatically across different urban neighborhoods.” This insight led to the introduction of flexible payment options that improved service accessibility while maintaining security.
Innovation at Apple
Currently, at Apple Ads, Kumar leads initiatives that merge machine learning with experimentation to optimize ad performance and marketplace efficiency. “Whether you’re predicting fire patterns or user behavior, the key is understanding the underlying patterns while respecting privacy and ethical considerations,” he says.
His team has introduced new methodologies in budget split testing and causal inference, significantly reducing bias in ad delivery while improving marketplace performance. The results have been impressive: enhanced ad relevance for over a billion users worldwide while maintaining strict privacy standards.
Looking Forward
As our interview concludes, Kumar reflects on his journey from Bokaro to Silicon Valley. “Data science isn’t just about algorithms and models,” he says. “It’s about understanding patterns that can make the world a little better, whether that’s preventing forest fires, providing financial opportunities, or improving urban mobility.”
He keeps a small coal piece from Bokaro on his desk, next to printouts of his NASA fire prediction maps. “These remind me that every data point tells a story. Our job is to listen to those stories and use them to make a difference.”
Kumar continues to mentor young data scientists, particularly those from underrepresented backgrounds. “The future of data science needs diverse perspectives,” he insists. The best solutions come when we combine technical expertise with varied life experiences and cultural understanding.
As he heads to his next meeting, Kumar pauses. “You know what’s amazing? That journey from tracking forest fires to optimizing ad marketplaces—it’s all connected by the same fundamental principle: using data to understand and improve the world around us. That’s what gets me excited to come to work every day.”
Looking ahead, Kumar remains passionate about pushing the boundaries of data science and machine learning. His vision extends beyond individual projects to the broader impact of data science on society. “The future is about collaboration across disciplines, industries, and cultures,” he says. The challenges we face—from climate change to financial inclusion—require us to think globally while acting precisely. That’s where data science can make its greatest contribution.”