Harsh Verma is a visionary data science leader with a passion for leveraging AI and machine learning to solve complex business problems. His career has spanned both academia and industry, where he has implemented innovative solutions that drive significant value. From his early days at Stevens Institute of Technology, Harsh forged key partnerships with Fortune 500 companies through the Business Intelligence Club, setting the stage for his transformative career in the corporate world. His work has had a profound impact on tech giants like Microsoft and e-commerce leaders like Chewy.
In this interview, Harsh shares his journey, insights, and vision for the future of data science and AI.
Harsh, your journey in data science is quite impressive. Could you share how it all began?
Certainly! My journey started during my time at Stevens Institute of Technology, where I focused on bridging the gap between academic knowledge and industry application. Through the Business Intelligence Club, I connected with industry leaders and honed my skills in machine learning and data science. This experience sparked my interest in how data could be leveraged to solve real-world business challenges.
My professional career took off when I joined Microsoft, where I worked on developing advanced Graph Neural Networks (GNNs) for online search advertising. I built a model that not only looked at individual user behavior but also considered the actions of similar users around them. This improved our ad targeting, resulting in a 5% increase in accuracy and a 6% rise in clicks. This early success motivated me to keep exploring new possibilities in data science and AI.
You’ve led transformative projects at both Microsoft and Chewy. What do you consider key to driving impactful results in data science?
The key lies in understanding both the technical aspects and the business goals. At Microsoft, for instance, my work with GNNs was not just about improving algorithms; it was about delivering value through better ad targeting for over 50 million users. By incorporating similar users data, we gained deeper insights into user behavior, allowing us to serve more relevant ads and enhance the user experience.
Similarly at Chewy, I tackled compliance issues by building a natural language processing (NLP) pipeline that automated the extraction of pricing data from compliance-related communications. This project saved the company over $500,000 in operational costs. It’s about using data science not just as a tool, but as a strategic asset to drive efficiency, revenue, and customer satisfaction.
What strategies do you use to identify and apply the right machine learning models for complex business problems?
My approach is always data-driven, focusing on both technical rigor and business impact. For example, I led a project to optimize pricing using machine learning models that adapted to real-time market conditions. The key was to build models capable of learning from the market and dynamically adjusting prices based on customer behavior. This strategic price adjustment resulted in a $20 million increase in annual profits.
I also emphasize the importance of exploring multiple models to find the best fit. In one of our projects, we tested different reinforcement learning (RL) models and ultimately chose a neural network-based model for its ability to predict continuous prices based on controlled reward input. This iterative process ensures that the chosen model not only meets technical requirements but also aligns with the business objectives.
How do you see emerging technologies like Generative AI and LLMs influencing data science applications?
Generative AI and Large Language Models (LLMs) are reshaping how we handle vast amounts of data. My current focus is on integrating these technologies into business applications to enhance productivity and customer experience. For example, LLMs can be used to filter and summarize information quickly, helping businesses navigate information overload efficiently.
Generative AI opens up new avenues for creating personalized user experiences and automating tasks that previously required manual effort. The potential is enormous, and I believe that by leveraging these technologies correctly, we can drive smarter decision-making and build more dynamic, user-centric products.
What advice would you give to aspiring data scientists looking to make a meaningful impact?
My advice is to remain curious and adopt an ownership mindset. In data science, it’s not enough to simply build models; you need to understand the ‘why’ behind each project and how it aligns with the larger business goals. Measuring outcomes is crucial – only what gets measured can be improved.
Additionally, I encourage data scientists to collaborate with cross-functional teams, including product managers, engineers, and business stakeholders. Understanding different perspectives leads to more robust solutions that can be implemented effectively.
Building a Legacy of Innovation
Harsh Verma’s journey from academic partnerships to leading transformative initiatives in AI and machine learning is a testament to his passion for data science. His work at Microsoft and Chewy has driven measurable business outcomes, from enhancing ad targeting accuracy to optimizing pricing strategies. As Harsh continues to explore the potential of Generative AI and LLMs, he remains focused on creating solutions that address complex business challenges and pave the way for future innovations in data science.
For those interested in connecting with Harsh Verma, you can reach out to him on LinkedIn: https://www.linkedin.com/in/hharsh/