Senior Manager Advanced Analytics
Australian Financial Crime Exchange
Total years of experience :13 years, 0 Months
(Fraud Bureau, Not for profit)
•Developed and launched a Fraud Search Engine for core members (Big 4 banks), replacing third-party tools and saving ~ AUD 70K annually on company cost.
•Brain and developer behind AFCXIQ©, a product for fraud checks for nonmembers after a successful MVP run. The estimated fraud prevention reduction is ~20%.
•Initiatives such as AFCX Bureau alerting system, an actionable detection suite for member banks across various fraud types and Graph technology-based intelligence exploration.
- North America, Data Science MVP to prevent Audit Failures of their 7700 sites throughout the United States.
Main Responsibilities
•Exploratory Data Analysis and descriptive analysis: Visualizations using python scripting.
•Feature Engineering using Spark Beyond® and open-source featuretools, Trane, tsfresh.
•Insights modelling for Customer Experience and Audit Failures: Developing context to generate hypothesis for insights using python visualization.
•Predictive modelling for audit failures, site level and standard level modelling: using Python-based frameworks, classical Machine Learning Algorithms (linear regression models GLMs, ensembles, bagging, boosting, stacking and deep learning using Keras)
•Hyperparameter tuning: using H2O AutoML
End to End Machine Learning Development Cycle
•Data Modelling - Key member in data model design, ETL scriptwriting in using SQL and python wrapper for Snowflake and AWS S3.
•Data Pipelines - Written using Luigi pipeline framework and Python
•Dashboards: Worked with UX/UI team for a different view of Audit failures, customer experience.
•ML Ops (model serving and productionization): Assisted the Data engineering team in hosting AWS EC2, using Linux, Luigi Pipelines for data engineering /model building.
•ML Ops (context update & scoring automation): Written scheduling jobs using Cronjob on Linux based AWS EC2 instance hosting the production environment.
Wholesale & Commercial Credit Risk)
•Key member of novel econometrics based EAD modelling resulting in estimated savings ~ 250K in economic capital.
•Introduced modern toolkits and machine learning-based approaches in the traditional quant environment, offering potential savings of two headcounts by automation.
data analysis for SME Corporate portfolio involving hierarchical clustering based insights, estimating missing values of features using probabilistic graphical models and gradient boosting techniques using Python 3.6
•Revamped the model monitoring framework for Property models using SAS EG, improving the execution efficiency from 1 day to 1 hour.
•Using SAS EG, created standard advanced graph plotting libraries, simplifying the process and better catering to the needs of different committees papers.
•Participated in comprehensive EAD modelling using Python 3.6. Replacing traditional CCF with new AUF. Beta mixture model as well as Bi-modal logistic regression with Expectation Maximization.
•LGD Modelling- trained models using statistical modelling - parametric Beta Regression, survival models such as Cox Proportional hazard and Accelerated Failure Times model with various distributions (Weibull, t, Chi2).
•LGD Modelling using machine learning approach - reinforced Polya urn learning.
•Dashboard Migration for Westpac Institutional Banking Portfolio using Python 3.6, Dash Library.
•R & D for early warning indicators framework establishment using principal component analysis based models based on Westpac share price volatility using Python 3.6 in Jupyter Labs and PySpark
Manager Credit Risk - Mortgages, Global Consumer Group
Citibank Australia | August
Australian Mortgages risk analytics unit for the highly complex global project ‘RRTS’ based on Fed and OCC (American Regulators) directives. Successfully reconciles the last 15 years BASEL and stress testing parameters and leads Australia to top position among APAC & EMEA teams.
Managed team of 3 subordinates:
•Business and Financial Models Development/5-year business plan formulation (risk-adjusted) for upcoming products of personal loan, credit cards, Auto Financing using M.S. Excel
•4Key member of X-Selling strategy development & customer segmentation modelling.
•Business Analysis - SAS Banking Intelligence Architecture Project.
•Responsible for the budgeting process of risk division.
•Leading with two subordinates from the retail risk side the migration of Legacy Cards to the New System. Participated in the migration strategy and designed the error-free formulation of the critical financial gaps.
•Revamped the existing obsolete credit data repository into a comprehensive consumer data mart incorporating risk and product team’s needs.
Managed team of 2 subordinates
•Implementation of highly urgent requirement of 1, 100 branches performance evaluation MIS packs before the target date of 1 month using SQL Server 2005 SSIS, SSRS and M.S. Excel.
•Consumer Data Repository designing and implementation using SQL Server 2005.
•Designed financial model for Risk-based repricing strategy, conducted risk-adjusted profitability and marginal segmentation calculation, break-even and scenario analysis, derivation of optimized rates using M.S. Excel.
in the Consumer Credit Data Mart Project implemented using IBM Cognos.
•Involved in outsourced Credit Scoring project and Limit Assignment Strategies using SAS BASE.
courses: AWS Certified Machine Learning Specialty (Exp