Apr 2024 – Nov 2024 · 7 mos

• As Head for Rakuten Institute of Technology India, managed a team of 20+ research scientists working across recommendation systems, customer intelligence, fraud detection, NLP, and computer vision for Rakuten’s global ecosystem of 70+ internet businesses.
• Delivered machine learning platforms and products generating tens of millions of dollars in incremental annual revenue through improvements in personalization, marketing efficiency, and trust & safety systems.
• Owned organizational planning and budget strategy, securing annual investment for the department while consistently demonstrating strong ROI through revenue and profit impact from deployed AI systems.
• Key initiatives:
• Customer Intelligence & Marketing AI – Led multi-year ML initiatives optimizing marketing and personalization across Rakuten’s ecosystem through platforms such as Customer DNA, Social Graph, Smart Segmentation, and Recommendation Systems.
• Trust & Safety / Anti-Fraud AI – Built large-scale ML systems to detect and prevent fraud across ads, affiliate marketing, reviews, and transactions, improving platform integrity across multiple Rakuten businesses.
• Language & NLP Platforms – Led initiatives leveraging unstructured text for address normalization, attribute extraction, catalog enrichment, and fraud detection.
• Academic Collaboration & Research – Established international research collaborations including work with Prof. Toyotaro Suzumura at the University of Tokyo, resulting in publications at top-tier conferences such as SIGIR.

Jan 2022 – Apr 2024 · 2 yrs 3 mos

• Leading multiple cross-geo teams covering customer understanding, customer marketing, social graph, fraud prevention, recommendations (https://rit.rakuten.com/research/customer/)

Jan 2020 – Dec 2021 · 1 yr 11 mos

• At Rakuten, I lead two teams across two departments (Research and Data Science):
• Customer Social Graph is a Rakuten research (RIT) project focused on leveraging graph models to predict relationships between users, products, Rakuten businesses in the absence of ground truth. Social Graph is used to improve Rakuten's customer understanding and for downstream tasks like marketing campaigns and recommendations across businesses such as Ichiba (ecommerce), Mobile and Fintech.
• Audience Sciences team uses causal modeling based framework for marketers for understanding business objectives and recommending which users to target, how to target (channel and Rakuten service) and when to target. Our current focus is on new user acquisition, reactivation, retention and churn prevention.

Feb 2018 – Jan 2020 · 1 yr 11 mos

• User Profiling and Personalization Platform
• Developing scalable ML algorithms and production pipelines to predict smartphone users’ demographics and interests based on app usage, geo location, browsing behavior, music/video usage.
• Using Profiles for personalization of Samsung apps such as MyGalaxy.
• Using look alike modeling for Customer Driven Marketing to drive Incremental smartphone sales.

Jul 2016 – Feb 2018 · 1 yr 7 mos

• Building User Profile Based on Smartphone App Usage
• Creating spatio-temporal profiles (home, work, frequent routes, etc.) for individual users using agglomerative clustering, sequence mining techniques.
• Predicting demographics related attributes such as income group, based on user's geo-location, app usage behavior, etc.
• Porting of algorithms to Apache Spark to achieve scalability(process millions of users).

Jan 2015 – Jul 2016 · 1 yr 6 mos · San Francisco Bay Area

• Designing machine learning algorithms for predicting customer behavior/usage of product using telemetry data.
• Implementation of distributed machine learning algorithms using Apache Spark for identifying frequent log pattern, co-relations between metrics and automated regression detection.
• Categorization of customer usage of product, based on storing and mining of terabytes of logs using an ElasticSearch cluster

Jun 2013 – Dec 2014 · 1 yr 6 mos · San Francisco Bay Area

• Performance Group (R&D) - vCenter Server Performance

May 2012 – Aug 2012 · 3 mos · Palo Alto, CA

• Designing and implementation of modules for storage, indexing and querying of performance data. Also built pattern recognition modules for detecting performance regression, outlier detection, step detection and correlation calculation.

Aug 2008 – May 2013 · 4 yrs 9 mos · Richardson, TX

• Predicting location of a user in an Online Social Network (OSN):
• Understanding the relationship geospatial proximity and friendship
• Instance based classification using k-nearest neighbors with variable depth
• Application of semi-supervised machine learning algorithms such as label propagation for predicting labels (locations).
• Identifying social cliques to predict most recent location of user.
• Other Projects:
• Developed TWinner, for predicting the news intent of the user and extract n words to enhance the location based user query for better search results.
• Developed MapIt, a GIS web application tool for identification and disambiguation of location of the apartment from unstructured ads.
• Software/Prototypes Developed:
• Vedanta: A powerful security analysis tool that builds a parallel profile (actual and psychological) for Twitter users by mining/predicting information such as location, age, interests, ethnicity, etc. and integrating several social networks such as LinkedIn, Google+, Foursquare.
• MapIt: A tool which displays the Craigslist apartment listings on Google maps. MapIt then integrates this functionality with the information collected from location based extraction of various web sources such as the city police blotter which makes apartment searching simpler and faster, helping the user to make a better decision.