Your Name

PROJECT - NEWZEALAND APPLICATION SCORECARD DEVELOPMENT

• Developed complex SQL and SAS queries to efficiently extract datasets exceeding one million records from Oracle databases.
• Conducted thorough analysis of Data Quality Reports to ensure data accuracy and reliability.
• Documented intricate technical dataflow, enhancing comprehension and facilitating future development.
• Refined and optimized legacy code to improve performance and maintainability.
• Extracted and meticulously prepared Application, Cross holding, and Performance datasets for analytical purposes.
• Performed extensive Vintage and Roll Rate Analysis to establish performance windows and define risk parameters.

PROJECT - OPTIMISE & VISUALLY ENHANCE THE MODEL DEVELOPMENT PROCESS USING R SHINY WEB APPLICATIONS

• Personalized and implemented variable reduction and binning techniques to optimize the feature reduction process.
• Decoded ANZ's clustering process from SAS, automated it in R, and integrated it into a web application using R Shiny.
• Reduced execution time by over 50% through code optimization for binning.
• Tested the tool on various datasets, ensuring timely resolution of bugs and errors.
• Developed intuitive graphs using ggplot and highcharter packages.
• Created a feature to automatically generate binning code in SAS, R, and Python.

PROJECT - UAT OF LOSS GIVEN DEFAULT (MORTGAGE AUSTRALIA PORTFOLIO)

• Conducted User Acceptance Testing (UAT) for LGD models for the Australian mortgage portfolio, ensuring regulatory compliance and adherence to internal policies, while collaborating with the Melbourne team to address discrepancies and data quality issues.
• Designed and executed test scenarios to validate the accuracy and efficiency of LGD models, developed UAT test scripts, and coordinated with business risk and technology teams to meet testing requirements and timelines.
• Provided regular progress updates to senior management, highlighting key issues, risks, and recommendations for model improvements, and demonstrated proficiency in applying the Cure Model methodology for better prediction of potential losses.

PROJECT - INTERACTIVE BINNING AND CLUSTERING APPLICATION IN PYTHON DASH

• Developed a Python Dash application to automate the binning and clustering processes, including functionality to read from and write data to Google Cloud Storage (GCS) blobs, ensuring efficient integration with Google Cloud Platform for seamless data management and storage.
• Integrated interactive charts and automated reporting features into the app, reducing modelers time and effort by 90% compared to traditional methods.
• Provided exclusive SAS, Python, and R code options within the app, allowing users to run the binning and clustering processes in various environments.
• Enabled export of datasets in Pickle, R, and SAS formats, facilitating further modeling activities and ensuring compatibility with different analytical tools.
• Refined the application to support multiple iterations of binning on the same variables, accommodating complex modeling requirements.
• Fulfilled all requirements of the modeling team by demonstrating the app to the Melbourne team, offering hands-on experience, and developing a notebook to replicate clustering functionality and stepwise regression in Python.
• Developed a detailed notebook to replicate clustering functionality, providing step-by-step guidance for implementing clustering algorithms in Python and ensuring accurate and replicable results.
• Created a comprehensive notebook for stepwise regression analysis in Python, offering clear instructions and code examples to perform and validate stepwise regression techniques.
• Designed the notebooks to enhance user understanding and ease of replication, including thorough explanations and practical examples to support the transition from SAS to Python for clustering and regression tasks.

PROJECT - BIAS ASSESSMENT AND MITIGATION IN MODELS USING MICROSOFT FAIRLEARN

• Assessed bias in models by evaluating sensitive variables ,using Microsoft Fairlearn to identify and address disparities.
• Explored and applied various bias mitigation methods, including pre-processing, in-processing, and post-processing techniques, to enhance fairness in model predictions.
• Developed interactive notebooks to facilitate users' understanding of bias mitigation strategies and their implementation.
• Tested bias mitigation methods on diverse datasets, ensuring effectiveness and robustness across different scenarios.
• Utilized exponential gradient descent processes to mitigate bias and improve fairness in the models.
• Conducted extensive quality assurance on the process, addressing and resolving various issues to ensure reliable and accurate bias mitigation.

SUPPORTING ACTIVITIES

• Generated and analyzed monthly model vetting reports, monitoring deviations in implemented models to ensure they are as expected.
• Utilized Power BI to visualize data quality issues, ensuring accurate and timely reporting for key stakeholders.