Jun 2022 – Mar 2023 · 9 mos

• Leading New User Trust Project
• User Trust Score Definition: Spearheaded the development of the trust score system for new users at the registration stage.
• A/B Testing & Experimentation: Conducted extensive A/B testing to evaluate the effectiveness of multiple models, analyzing their impact on critical metrics such as signup completion rate, fraud incidents per signup, new user registration rate, and bookings from new users.
• Model Development & Optimization: Developed and refined machine learning models to accurately generate user trust scores. The models leveraged a wide range of features, including user login session data, inputs from external partners (e.g., mobile verification, IP details), and user-submitted information. Achieved a recall of 81.2% and a precision of 20.7% on test data.

Jan 2021 – Jun 2022 · 1 yr 5 mos · Hyderabad, Telangana, India

• I am working as a Senior Data Science Manager at WDX India. My team is responsible for Accessibility products (VoiceAccess, Narrator) and sharing platform (Windows Share)
• Voice Access: Designed real time next action prediction model for "VoiceAccess" feature that works offline with Win 11. Combined the model predictions with speech-to-text converter for improving efficiency by 45%.

Jun 2019 – Jan 2021 · 1 yr 7 mos · Bengaluru, Karnataka, India

• Address Parser: Developed the nation’s 1st NLP model for classifying Indian addresses as deliverable/ non-deliverable based on various factors (i.e. Monkey-typing, locality, society/ last mile identifier, point of interest). Patent Filing in process
• RTO Prediction: Designed & deployed RTO (return to origin) prediction model using 300+ features (customer behavior, past transactions, address features). Achieved precision – 0.78 & recall – 0.52 with the final ensemble of model
• Fraud Transaction Detection: Designed fraud detection model for international transactions on Indian merchants using the Razorpay PG. The model is run in real-time for all the international transactions and the final decision is a combination of model results and custom rules.

Oct 2018 – Sep 2023 · 4 yrs 11 mos · New Delhi Area, India

• Projects:
• AdScout AI: Developed & deployed a deep learning recommendation tool that identifies optimal advertising opportunities for soccer players, tailored to their career trajectories, preferences, and emerging market trends. The model leverages over 500 unique signals from more than 20 diverse sources to provide personalized brand recommendations, enhancing sponsorship and endorsement potential for athletes.
• ProInsight AI: Designed and deployed an XGBoost model to forecast injury risks for soccer players in the upcoming season. The model extracted and analyzed over 1,000 features from various data points, including health records, match schedules, playtime patterns, coaching influences, and training data sourced from Apple Watch, soccer clubs and ESPN. Achieved a precision of 80% and recall of 65% across two years of EPL League 1 player data, providing actionable insights for injury prevention and player management.

Oct 2015 – Sep 2018 · 2 yrs 11 mos

• Product Matching : Created sequential model using 2 CNNs & 1 NLP for Product Matching across various E-commerce websites. Achieved accuracy 99.5% across Sears, K-Mart, Target, Amazon, and Nordstrom website and 95%+ across others
• Product Recommendation : Implemented CNN for feature extractions to enhance the existing Collaborative Filtering based product recommendation. Improved the F1 score by 60%
• Machine/ Deep Learning Templates : Developed 14 templates for machine learning algorithms. Various functions such as one-hot encoding, train-test split, outlier detection, normalization, performance recording against standard metrics are automated.

Aug 2012 – Sep 2015 · 3 yrs 1 mo · Greater Delhi Area

• Sookshm is a consulting organization that works with medium-size and smaller manufacturing companies to help them compete and grow profitably. In the 1st year alone Sookshm has helped over 20 Small Medium Enterprises gain access to strategies that increase efficiency and promote growth. Resulting from the organization’s work, Sookshm's clients have realized a total positive economic impact of more than $ 2 million through cost reduction.