Mar 2023 – Feb 2024 · 11 mos · Bengaluru, Karnataka, India · Hybrid

• Leading a team of 4 data scientists and SMEs to build anti counterfeit solutions in the packaging industry using classical computer vision and AI.
• Trained and deployed YOLOV5 object localization model on the app making product discovery 60X faster.
• Evaluated and executed diverse image quality systems (FR-IQA and NR-IQA) to enhance camera image quality

Aug 2019 – Jan 2023 · 3 yrs 5 mos · Bengaluru Area, India · On-site

• Leading Food-Menu DS charter to increase the relevance of menu. Led 5+ successful DS experiments impacting scorecard metrics M2O and AOV.
• Realtime LTR based recommendation model improving conversion by 0.08pp and AOV by 0.60. Composite features to save on expensive Cx-item features.
• New cross sell widget designed and integrated in menu leading to 20% widget CTR and 9% order share.
• Built real-time GBTRegression ETA service model with mean absolute error reduced from 6.3min to 4.4min.
• Image processing based automation tool using inpainting model for Swiggy designers to aid banners generation saving TAT from 7 to 3 days.

Mar 2018 – Jul 2019 · 1 yr 4 mos · Bengaluru Area, India · On-site

• Ideated and developed modified U-Net architecture for segmentation of optic disc and cup in Glaucoma screening. Achieved state-of-art performance on open-source dataset for cup disc segmentation. Work published in ISBI 2019.
• Localization of Abnormalities using Class Activation Maps in OCT scan to detect Macular Edema. Used VGG-16 as base architecture with modifications to classify as well as localize abnormalities in OCT scan.
• Smart Annotation Strategy for prioritization of labelling of unlabelled data. 4% increase in F1 score on test data using strategy compared to random selection of data for annotation.

Jun 2015 – Mar 2018 · 2 yrs 9 mos · Bengaluru, Karnataka, India · On-site

• Streamlined Customer Centric Recommendation Engine for improvised customer engagement & credit mix
• Idea-ted and Built faster and scalable solution using concept of Collaborative Filtering and Association Rules
• Built solution in Map Reduce framework using Java and Hive.
• Developed Customer Segmentation framework to identify engagement of customers with Citi credit card
• Designed multiple campaigns to target customers based on historical performance yielding high incremental sales
• Developed framework to re-engage customers by proactively targeting them with personalized offers.

May 2013 – Jul 2013 · 2 mos · Hyderabad Area, India

• Worked on increasing Fault Tolerance in Database Server.

May 2012 – Jul 2012 · 2 mos · New Delhi Area, India