As a scientist at Wayfair, Sachin Chhabra is helping fulfill the company’s mission of enabling customers to create their own sense of home. To this end, Chhabra utilizes state of the art machine learning techniques to surface relevant product recommendations to customers.
As part of the search and recommendation systems team at Wayfair, Chhabra develops machine learning models that utilize deep graph neural networks (GNN). In recent years, GNNs have become increasingly popular as businesses go beyond developing models for two dimensional and three dimensional signals such as images and video to learning from structured data embedded in graphs.
“We have developed a bipartite GNN with users and items on each of the axes,” says Chhabra. “Every product has a vector representation, and we utilize GNNs to compress that information in an embedding space. We can then effectively measure the distance between items to surface relevant recommendations for customers.”
Once Chhabra’s models are deployed, the machine learning models will help tens of millions of customers satisfy their home-related needs. Chhabra says that the opportunity he has been given to make such a large impact is remarkable given that he is an intern at Wayfair.
“I dont know of many places where interns are able to take on projects that are so important to the company,” says Chhabra.
Chhabra is currently pursuing a doctorate in machine learning at Arizona State University. His research is focused on computer vision. Prior to Wayfair, Chhabra had never worked with GNNs.
However, taking on projects that are outside his comfort zone has been a running theme throughout Chhabra’s life. Like his parents and brother who are doctors, Chhabra found from a young age that he was naturally drawn to math and science. Chhabra challenged himself by taking part in India’s highly competitive math Olympiads, where he placed in the top twenty students in the country.
Chhabra completed his bachelors at the Vellore Institute of Technology, after which he was admitted to Arizona State University to pursue a masters degree. He ran into a fellow student from his alma mater, who told him about machine learning. Chhabra was intrigued, and enrolled for advanced courses in natural language processing and computer vision.
He found the going tough initially because he had no background in machine learning. However, he pushed himself to become familiar with the course material. He completed the courses at the top of his class.
Chhabra was intrigued by the possibilities opened up by machine learning – so much so that he decided to pursue a doctorate in the field. Today, his research focuses on developing machine learning models that use transfer learning. Transfer learning involves training a model to learn a new task by drawing on knowledge that has already been learned.
Wayfair is Chhabra’s second internship – previously, he worked for the Systems Imagination company, developing pretrained models to predict COVID-related outcomes in counties across America.
In this article Chhabra outlines three reasons as to why science doctorate students should seriously consider pursuing an internship at Wayfair.
Opportunity to have a real-world impact
Chhabra says that while working at Wayfair, interns are given the opportunity to have complete ownership over projects that make a tangible difference in the lives of millions of customers around the world.
This comes with a sense of provide and accomplishment. In addition, Chhabra says the real-world repercussions of his work keep him laser focused on delivering improvements, however small they might seem.
“There was a time when the performance of my model was one percent off what was projected,”says Chhabra. “This would be perfectly acceptable in the world of academia. However, my manager told me that even a shortfall of one percent could potentially have financial repercussions of millions of dollars. That was a huge ‘aha’ moment for me. Now, I am always looking to drive improvements, no matter how small they might seem. This has made me a better scientist in so many important ways.”
Flexibility in the choice of assignments
Chhabra says that companies often hire interns to complete tasks that are clearly pre-defined, and often aligned with a short-term objective for the company.
“That’s not the case at Wayfair,” he says. “Here I was given complete freedom in what I want to do.”
Chabbra says that when he joined the company, he had a deep conversation with his manager and mentors about his areas of interest. Wayfair suggested that Chhabra build a recommendation system using GNNs.
“Wayfair gave me complete flexibility in how to solve the problem - from designing the architecture, to feature selection and evaluation criteria,” says Chhabra. My manager was aware that I didn’t have experience in GNNs. So he planned time that was devoted exclusively to my studying the latest research, and coming up to speed. This took some time as I am the kind of person who likes to understand everything from the inside out. But my team trusted that I would eventually be able to come on board and deliver results.”
Opportunity to work with state-of-the-art advancements in machine learning
Chhabra says that Wayfair’s sheer scale directly translates into a wide ranging variety of problems that can only be solved by machine learning. He says that the team is constantly experimenting with state-of-the-art techniques to deliver quality recommendations to customers.
“We use transformers for sequence-to-sequence modeling and text data processing,” says Chhabra, referring to deep learning models that can weigh the importance of each part of the input data to evaluate context.
“We are conducting groundbreaking work we are doing in areas like computer vision, GNNs and semantic search. The team has one weekly session, where scientists have an opportunity to present their work and get ideas from their colleagues. We also have monthly meetings that are exclusively dedicated to discussing the latest research.”
Chabbra says that his manager and mentors have encouraged team members to publish scientific papers based on their work.
“My journey at Wayfair has been much more than working on one project,” Chhabra says. “I feel like the team at Wayfair is really invested in my career. It’s not an understatement to say that Wayfair is an incredible place for science interns.”
