• Built predictive model to rank potential customers in order of likeliness to apply for insurance to improve targeted marketingefficiency and information gathering time.• Leveraged LLMs (Azure GPT-3.5) to summarize unstructured applicant information to optimize insurance applicationreview process.

Analyzed how linguistics influence persuasive power of textual content in the context of customer reviews and fake news.

• Led the effort to design a tool to parse unstructured text data in PDFs and extract relevant information leading to aperformance improvement of over 40%.• Spearheaded user acceptance testing (UAT) to ensure smooth roll-out of new features.• Conducted code reviews to ensure code developed by team was up to standard and provided feedback where required.• Built an information retrieval system to retrieve relevant PDF documents based on question asked by users.• Collaborated with end users and other stakeholders to design and implement a continual model training pipeline based onuser interaction with the tool to incrementally enhance model accuracy over time.

• Developed machine learning models for text classification and text similarity using transformer-based models like BERT.• Trained object detection models to extract relevant graphical information from images with over 90% accuracy.• Deployed and maintained machine learning models as REST APIs in containers on AWS EC2.

• Optimized the speed of a Geocoding algorithm by more than 20% by developing algorithmic modifications to identify pointsof interests from a batch of location data using statistical analysis.• Collaborated with other data scientists to build a demand forecasting model for a large Indian online fashion retailer topredict future orders using time series forecasting.• Created an outlier and error detection system for user entered data using clustering algorithms and suggest possiblecorrections.