Naveena Koneru

• Analyzed large datasets to identify trends and patterns in customer behavior.
• Built predictive models using machine learning algorithms for data analysis.

Product: Insights - Counterfactuals Module

• Research and Development of a Counterfactual API Module with the DiCE library for Loan Applications.
• Hyperparameter tuning for various ML models.
• Identified key factors that an individual loan applicant must change to get a loan approved.

Product: Insights - Knowledge Graphs Module

• Research and development of causal knowledge graphs with a user interface.
• Aggregated root cause analysis, with the user’s expert knowledge of domain parameters and constraints.
• Created a weighted directed acyclic graph (DAG) of the most important features that contribute to the loan application decision.
• Tested the CausalNex and PyRCA libraries, as well as the PC algorithm.

Product: Prompt Engineering for QA Automation with Azure OpenAI

• Used GPT3 and GPT4 to generate textual and code-based test cases with allowed inputs to be either code following Pep8 standards or business requirement documents

Product: Fuzzy Logic Matching Module

• Developed two different algorithms tailored to identify types of bank transactions: self-transfers or transfers between related companies from the transaction narration using Python.
• Increased both accuracy and precision from 40% to more than 85%.
• Integration into the Early Warning System.

Product: Early Warning System

• Analyzed and processed data as necessary, both for individual use cases and overall business requirements.
• Developed use case logic and oversaw implementation in development, UAT, and production stages.
• Communicated and discussed the results of the output with the client, project managers, and data engineers.
• Was responsible for maintaining the full use-case logic code base, including modifications of existing code, and integrations required.