Since its debut in 2011, Mumbai-based Pepperfry has been a game-changer in the way Indians shop for furniture and stylish home decor. Pepperfry was one of the early adopters of cutting-edge technology when it launched its online marketplace. Apart from classy furniture, Pepperfry is known for its wide range of furnishings, lighting, and other home utility products. Started by former eBay executives Ambareesh Murty and Ashish Shah, Pepperfry caters to every taste and requirement for every type of home.
AIM got in touch with Devvrat Arya, vice president of technology, at Pepperfry to understand how the furniture giant is implementing data science and who are cut out for such roles.
“We are at the cusp of a new era where AI will become ubiquitous, and machine learning and deep learning will power everything. To succeed in this field, one must have a strong foundation in linear algebra and statistical analysis, as every algorithm is based on these fundamental concepts,” said Arya.
Pepperfry’s AI & Analytics Play
Pepperfry’s data science team is fairly small and focuses on addressing problems related to the company’s customers and business. The organisation strongly believes in the super-lean methodology and begins by defining problem statements before seeking individuals to build a team around the project. This approach also governs the team’s structure and hiring process, which align with the super-lean methodology.
Pepperfry implements AI/ML models to improve the customer experience and one specific area that they focus on is anomaly detection. The data science team is analysing historical order-placed patterns from the past year and examining all possible combinations. By using various ML algorithms, the team can identify order anomalies that deviate from the predicted order-placed pattern logic. Once an anomaly is detected, the ML engine automatically notifies the relevant parties of the possible root cause of the issue. The data science team and developers collaborate to address the problem and prevent any technology leaks that could negatively impact business outcomes.
Although in the beta stage, Pepperfry has also developed a visual search engine that allows customers to upload an image of furniture they’re interested in and receive recommendations for similar products. The algorithm compares the uploaded image with database images to find the top similar products. This feature simplifies the process of searching for furniture and decor items and provides insight into customer preferences for a more personalised shopping experience.
Read more: Data Science Hiring Process at Pepperfry
Interview Process
Pepperfry’s interview process for hiring data scientists involves several stages. First, the candidates are given an initial assessment consisting of complex and randomised data science and Python-based questions to assess their skills and knowledge. The first interview round evaluates the candidate’s analytical skills and suitability for the role based on personality traits. In the second technical interview round, the focus is on testing the candidate’s knowledge and understanding of multiple machine learning and deep learning algorithms. Lastly, there is an HR round for negotiating the salary and the issuance of an offer letter.
Skill Set Required
Pepperfry emphasises the importance of a candidate’s personality fitting the role they are applying for. The company’s initial evaluation of potential candidates focuses on several personality traits, including analytical skills, agility, communication abilities, honesty, curiosity, and strong work ethics.
The data science team focuses on visual-based customer experiences and has prioritised the use of deep learning-based libraries and frameworks. Tensorflow, Pytorch, and Scikit-learn are the primary tools utilised for data analysis and deep learning projects. Additionally, Fast RCNN and Yolo libraries are extensively used for object detection and segmentation, and SWIN transformers are employed for efficient image-based result classification. The company intends to delve deeper into GPT to increase organic traffic gradually.
Applicants with a strong background in linear algebra, probability theory, statistical analysis, and distribution will be viewed as an asset.
Arya, who has interviewed over 500 data science engineers, has noticed that the most common mistake candidates make is by attempting to apply a simplistic statistical and analytical approach to the ML process. This, he says, is inadequate as the disparity between statistical and ML analysis is significant, and a uniform strategy is unlikely to be effective in solving the problems at hand.
Work Culture
“Pepperfry, as an organisation, values a culture that is diverse and inclusive and promotes positivity,” said Arya. Here, about 35% of employees are females against 65% males. Despite being in existence for over eleven years, they still maintain a start-up mentality through continuous innovation while remaining focused on its business goals. The company employs a flat organisational structure, which eliminates communication barriers between colleagues.
“We have abolished the cabin culture this year, and now all employees sit together regardless of their role for faster and easier ideation, discussions, and conflict resolution within the organisation,” he added.
Pepperfry offers various perks to its employees, such as a hybrid working model, the flexibility to choose working hours, and ensuring a healthy work-life balance. The company consistently invests in its employees to enhance their skills and expertise in the latest technology and tools and provides access to top-notch learning resources and certification programs.
“If you have the knack for solving complex and innovative problems with the right mix of startup energy, Pepperfry is the place for you,” concluded Arya.
Click here to apply.
Read more: Data Science Hiring Process at Livspace
