Jan 2025 – Nov 2025 · 10 mos · Greater Bengaluru Area · Hybrid

• Automated QC for catalog

Sep 2023 – Dec 2024 · 1 yr 3 mos · Bengaluru, Karnataka, India · Hybrid

• Built Trippy (AI-powered Chatbot for trip planning)
• Hotel's Ranking

May 2022 – Aug 2023 · 1 yr 3 mos · Bengaluru, Karnataka, India · On-site

• Skin Care and Hair Care personalised recommendations : launched V0 DS solution that increased use engagement of beauty products from 3% to 10% in 4 months
• Generalised personalised recommendation framework : designed to eliminate additional DS and DSE effort for custom product racks
• AI Stylist, a salient feature for Fashion E-commerce: Increased coverage to 90%+ of live styles and products bought in last 2 years with average coherence of 3.5 (out of 5). Project was highlighted in media by company’s PR team and was monitored at CXO level
• Designed for data collection to train AI Stylist V1 model, with collaboration with DSE
• Generated salient samples for validation and curation to augment AI Stylist training data
• Product embedding: Fine-tuned RoBERTA to generate universal product embeddings that are used in all new recommendations projects
• Myntra Studio and Myntra minis - Trained DL model for recommendations, increasing both CTR (by 20%) and coverage (to 90%+) on product display pages
• Shop from banner : Hackathon project
• Collaborated extensively with Data Science Engineering, ML Platform, Analytics, Product Management as well as Engineering teams to plan, execute and deploy multiple projects end-to-end

Jan 2019 – Apr 2022 · 3 yrs 3 mos · On-site

• As first hire of international Audience Targeting team, mentored and familiarised later hires with project-specific trainings
• Increased coverage of Audience Targeting product from ~ 50-100 segments in only 3 English markets, to 350+ segments across 80+ countries and 17+ languages, with minimum of 10% CVR again and positive CTR gain and impressions share
• Worked extensively with US counterparts for bidirectional technology transfer as well as joint experimentations
• Mentored intern in very first year to create a sanity check module that monitored daily pipelines and effectively eliminated production bugs
• Mentored interns to create multilingual multilevel classifier, leveraging XLM-R and Unicoder, covering 17+ languages and had accuracy of 80%+
• Explored creation of seasonality based segments and repeat-buy segments
• Expanded Similar Audience product to non-English 50+ markets covering 20+ languages
• Expanded Travel and Seasonal segments to 4 international markets, covering 3 languages (English, French, German)
• Worked with multiple algorithms - LR, XGBoost, CNN and LSTM with simple attention, Hierarchical graph networks with graph networks, Locality based Hashing, K-Means clustering
• Worked extensively on data preparation modules in SCOPE (U-SQL) on CosmosDB

Jul 2018 – Jan 2019 · 6 mos · On-site

• User intent modeling using Bing search queries and user activity analysis on various Microsoft online properties for targeted ads on Microsoft Universal Storefront
• As first data science practitioner of Business Applications Group, pioneered use of ML and Data Science

Jul 2015 – Jul 2016 · 1 yr · Bengaluru Area, India

• Health and Medical Equipment/Magnetic Resonance Imaging
• Assisted on the Heart and MR Spectroscopy before getting responsibility for two Perfusion MRI modules. On the way, picked up on Qt, MVC (Model-View-Controller) architecture and Cuda as required for the job. Towards the end, helped begin an effort to establish a common UI platform for cross-model uniformity.

May 2014 – Jul 2014 · 2 mos · Hyderabad Area, India · On-site

• VLSI algorithm designs for partitioning and floorplanning. Developed JAVA library for graph and hypergraph algorithms.