BALKRISHNA JOSHI

Auto Insurance Quote Generation – Functional & Automation Testing:

Project Overview:
A customer-facing web application that handles automated email and SMS communications, such as payment failure notifications and enrollment messages. Ensured the validation of email/SMS content, links, PDFs, customer details, and responses, while updating customer information in the database and providing status updates to upstream systems.

Key Responsibilities:

• Collaborated with business and technical teams to define testing requirements for email and SMS workflows.
• Developed test strategies& requirements, test plans, and test cases for functional and API testing.
• Managed end-to-end Email, SMS & PCF testing, ensuring the accurate validation of communications under diverse conditions.
• Automated email flow using Eclipse and Java and conducted API testing with SoapUI and Postman for seamless integration with JMS and Kafka.
• Validated system performance and monitored logs using Tibco and Datadog to identify and resolve issues.
• Actively participated in Agile ceremonies, including Sprint Planning, Backlog Refinement, Daily Standup, and Sprint Retrospectives, ensuring test automation tasks align with sprint goals.
• Analyzed defects, tracked issues, and collaborated with development teams to ensure timely resolutions.

Insurance Policy Management System – Web & API Automation:

Project Overview: I was responsible for ensuring the accuracy and reliability of Allstate’s Insurance Policy Management System, including policy creation, updates, renewals, and cancellations. I performed both manual testing and automation to validate system functionality and ensure seamless operation.

Key Responsibilities:

• Performed policy creation, endorsements, and renewals, ensuring accurate updates across states and company codes.
• Designed and executed API tests using REST Assured, SoapUI and Postman, validating policy services and ensuring data consistency.
• Developed and maintained automation scripts for functional testing using Python and LWF frameworks for scalability.
• Integrated and executed automation tests within the Jenkins pipeline, enabling continuous testing after every build, ensuring fast identification of issues.
• Managed Jenkins pipeline runs as part of the CI/CD process, troubleshooting and resolving failed test cases in IntelliJ.
• Collaborated with development teams to log and track defects in JIRA, ensuring timely resolution and adherence to project timelines.
• Weekly regression suite run in order to ensure no defect leaks, Reduce Defects reported in tracker.

Auto Insurance Quote Generation – Functional & Automation Testing:

Project Overview:
For the Auto Insurance Quote Generation project, I ensured the accuracy and functionality of the web-based quoting system. This system allowed customers to generate customized insurance quotes based on factors like car model, driving history, and location. My focus was on both functional and non-functional testing to guarantee a seamless user experience.

Key Responsibilities:

• Designed and executed test cases to validate the quoting system, focusing on input validation, accurate premium calculations, and UI usability to ensure an optimal user experience.
• Conducted both functional and non-functional testing, ensuring error-free quote generation and seamless interaction between system components.
• Automated test scenarios for quote generation using Selenium Web Driver (Java, TestNG), covering various edge cases, input validation, and business rule compliance.
• Performed tests using Selenium Grid to ensure consistent performance and compatibility across Chrome, Firefox, and Internet Explorer.
• Verified premium data accuracy in the database through SQL queries and validated the correct retrieval of customer data using REST APIs.
• Logged and tracked defects in JIRA, collaborating with development teams for prompt issue resolution to maintain project timelines.
• Integrated automated tests into the Jenkins pipeline to enable continuous execution and efficient test management post-build.

Automated Underwriting System – Machine Learning

Project Overview:
Developed an Automated Underwriting System to streamline the insurance application process by automating risk assessment and decision-making, improving efficiency and accuracy.

Key Responsibilities:

• Acquisition of usable data from valuable data sources and performed Exploratory data analysis to analyze and investigate data.
• Evaluated various machine learning algorithms including Logistic Regression, Decision Trees, Random Forests, and Gradient Boosting, XGBoost for classification tasks.
• Implemented a ensemble techniques to combine the predictions of multiple models, improving overall accuracy and robustness.
• Implemented hyperparameter tuning, cross validation to optimize the accuracy and better performance.
• Successfully reduced underwriting processing times by 55% while maintaining high accuracy of 94% and precision of 89%.
• Demonstrate and represent the outcomes in clear manner to Business owners.

Sentiment Analysis & NER model – NLP

Project Overview:
Developed a sentiment analysis model using advanced NLP techniques to classify text data as positive or negative. Designed and implemented a Named Entity Recognition (NER) pipeline to extract key entities from unstructured insurance documents, leveraging both traditional NLP methods and transformer-based models for enhanced accuracy and performance.

Key Responsibilities:

• Designed a sentiment analysis system for customer feedback, initially tested traditional NLP approaches before transitioning to RNN and LSTM models for improved accuracy.
• Increased model accuracy from 78% to 90% through hyperparameter tuning and cross-validation, deploying the model via Flask API to analyze over 50,000 customer reviews monthly.
• Implemented TFIDF, Word2Vec, and count vectorizer to improve contextual understanding through numerical text representation.
• Developed an NER pipeline for unstructured insurance documents using NLP techniques and transformer models like BERT.
• Conducted data preprocessing with tokenization, lemmatization, and custom regex, establishing baseline performance with traditional methods
• Improved F1-score from 75% to 92% through hyperparameter tuning and optimization techniques, automating data extraction to reduce manual effort by 60% and enhance model performance.