Sep 2023 - Present

Data Scientist at Canada Life

● Architected and deployed Smoker Propensity Model using XGBoost and advanced feature engineering on Azure ML, improving underwriting recall by 20% and achieving 310% lift over baseline, delivering around 1,500 hours in annual automation savings ● Led LLM-powered feature extraction initiative using OpenAI APIs and prompt engineering to process 386K historical medical records, expanding usable training data by 96% with less than 4% error rate through data wrangling and integration ● Designed LLM Judge validation system for automated quality assessment of extracted features, partnering with underwriting teams to reduce manual validation workload while maintaining model reliability ● Established end-to-end MLOps infrastructure with MLflow, Docker, and CI/CD frameworks across FIT/UAT/PROD envi

Apr 2021 - Aug 2022

Data Scientist at Eaton India Innovation Centre

● Developed a change point detection algorithm for predictive maintenance of UPS devices using statistical analysis on 600+ UPS units with 20 parameters, deployed via Power BI dashboards and presented insights to operations teams ● Architected serverless monitoring solution on Azure Functions, reducing deployment time from 4 hours to 20 minutes while enabling real-time parameter tracking for critical infrastructure across multiple data centers

Aug 2019 - Apr 2021

Data Scientist at Hitachi Consulting

● Architected an end-to-end MLOps system using MLflow and Seldon-core for automated Kubernetes model deployment, reducing deployment time by 60% and eliminating DevOps dependencies for data science teams ● Built an NLP solution using AWS Textract, NLTK, and SpaCy for structured data extraction from transport documents, achieving 80% accuracy in automated field identification and entity recognition ● Developed ensemble models (XGBoost, LightGBM) on a 5GB manufacturing dataset, achieving 78% f1-score for tire defect prediction with SHAP-based interpretability for process optimization

Jul 2018 - Jul 2019

Associate Innovation Engineer at Zensar Technologies

● Developed CNN-based computer vision models for autonomous navigation and scene segmentation using 50,000+ curated images, achieving 78% steering prediction accuracy for a self-driving vehicle prototype ● Deployed production-grade facial recognition system achieving 79% accuracy for 40+ employees and footfall detection achieving 89% accuracy, processing 1,100+ concurrent requests per second via optimized Django REST APIs for business intelligence analytics