Aug 2018 – Jul 2021 · 2 yrs 11 mos · Bengaluru, Karnataka, India

• Org Building
• Responsible for hiring Data Scientists for Flipkart by conducting coding, machine learning design
• interviews.
• Grew the data science team in User Acquistion and Retention team by hiring, mentoring data scientists
• for over 3 years. Mentored other data scientists on projects like review ranking, size and fit and
• machine translation.
• ANSWERING PRODUCT QUESTIONS USING USER GENERATED CONTENT:
• Auto Answering was a critical project for Flipkart since users who usually posted questions had an
• high intent on buying a product.
• I developed machine learning models for automatically answering answers using UGC content like
• reviews and Answered Product Questions. Proposed a method to get simple and hard negatives for
• the models and thus reduced the need for manual annotations.
• The models developed had precision of 56% beating the baseline models by over 10%.
• ANSWERING PRODUCT QUESTIONS USING PRODUCT SPECIFICATIONS:
• I developed models with researchers from IIT KGP to answer product questions posted by users
• using product specifications.
• The models developed were presented as a paper in ECNLP conference.
• DETECTING SEARCH TAGS IN PRODUCT QUESTION ANSWERS (PQA):
• I developed topic models (optimised for short text and used word embeddings learnt on Flipkart
• data) to identify tags for questions posted by users on product pages.
• The models developed were deployed for the beauty verticals.
• The models developed were used by the product catalog team to identify missing product
• specifications for categories.
• IN HOUSE MODELS FOR REVIEW AND QUESTION MODERATION:
• Developed in-house models for identifying reviews and questions that did not meet Flipkart's
• moderation standards.
• Saved more than 2 million dollars in costs since it reduced effort for human moderation of every
• review and question

Aug 2017 – Aug 2018 · 1 yr · Bangalore

Aug 2015 – Aug 2017 · 2 yrs · Bangalore