AI & Financial Standards SpecialistProject: Development of an IFRS-Focused RAG Bot • Spearheaded the development of an innovative Retrieve and Generate (RAG) bot, designed to provide automated, first-level policy support in alignment with International Financial Reporting Standards (IFRS). • Engineered the bot to accurately retrieve relevant IFRS policy data, ensuring it generates context-aware responses tailored to the unique queries of financial experts and auditors. • Collaborated closely with a team of AI technologists and finance professionals to integrate advanced machine learning algorithms, ensuring high accuracy and real-time policy updates. • Played a pivotal role in the bot’s design and functionality, focusing on user experience to create an intuitive, efficient tool for policy interpretation and application. • Successfully enhanced decision-making efficiency for financial experts, significantly reducing response time and improving accuracy in policy application. • Contributed to ongoing improvements based on user feedback, demonstrating commitment to technological innovation and excellence in the field of financial reporting.

Was part of the NextGen program in a major bank based out of the United Kingdom. The objective of the NextGen Program was to augment the Internal Auditors with Process Mining which would further result in saving the number of hours per audit. As part of the program we as a team looked at Service Delivery Payments Audit for High Value Payments which runs on Global Processing Services while also being dependent on other systems such as SWIFT messaging gateway, WOLF sanctions screening etc. A series of Analytical test were built using ETL pipelines(in SQL) and Celonis Process Mining to flag exceptional payments which did not adhere to the typically expected audit trail. For example Segregation of Duties check, Payment Cover Matching, Payment Exceptional Handling, Fraud Pattern Queue monitoring and successfully pointing out payments not subjected to a manual callback.Resulted in Smarter Sample testing for the auditors and hence reducing the number hours spent on the audit by 20%

Time-sheet Analysis -: The objective of the project was to automate the collection and aggregation of all time-sheet data for Internal Audit Employees on a weekly basis. Developing dashboards and reports for several use cases including, training information to analyze where training budget hours are being consumed. Time-sheet timelines to understand and highlight areas/individuals who are completing time-sheets in an untimely manner. Time-sheet Code analysis to show where each user is spending most of their time. Automate constant monitoring of overall time-sheet metrics available from the time-sheet input such as Overtime, Under-time etc.Blueprint Analysis -: Automate the monitoring of key Blueprint metrics for Internal Audit employees on a monthly-basis. Develop reports and dashboards for several different use cases across the department, for example: Professional Years of Experience. Certifications, Core Competency levels, Country Manager Profiles, Leadership Team, Auditor in Charge, Audit Delivery etc.Performance Management -: On an annual basis the Client team conducts a thorough Performance assessment of its employees where the process is subjective as it depends on the views of the respective MD/VP and Reporting manager. With the aim of minimizing subjectivity and making the whole process more inclusive and objective, the Client team aims to develop a robust Decision Tree Regressor which would in fact look at the criteria of performance evaluation with a more objective view.Talent Management -:The Client carries out an exercise of identifying the HiPots (High Potential) Employees in their workforce so that they can form the Target Segment Employees when it comes to assessing Learning and Development practices and Reward and Recognition practices. The objective of this module was to ensure that the High Potential Employees were identified with a more robust approach.

Saudi ArabiaDeployed as a Data Scientist in the Credit Policy and Control Department for a major telecom provider in Riyadh, where as a Data Scientist we are involved in the automated control monitoring of the following Reports:Whole Sale Aging Report: On a quarterly basis, the AR department sends a list of International Operators which have partnered with the client to provide service or sell products across three market segments, namely RC, Roaming, and Traffic. As a data scientist, in the credit policy and control department we have successfully automated the monitoring of the operators whether subsidiary, on hold or any other. As part of the report we monitor the aging amounts of each operator classified into 4 or more buckets and help the client keep track of the Aging amounts that are becoming red flags and need corrective action.Enterprise Business Unit Reports:a. Dunning Exclusion: As part of the Report, we monitor and help client gather justifications from the business for the Enterprise unit customers which have been excluded from dunning, yet showcase suspicious behavior in terms of exceeding the credit line or irregular billing history or bank guarantee expiry.b.  Excess Credit limit (Billing account and Company level): In this report we tend to monitor any and all revisions in enterprise customer's credit line via an SR Request raised by the customer. As a data scientist, we have successfully helped in creating dashboards that can provide actionable insights into the revisions of credit line which have violated the Sales CAP credit limit as per the credit policy.c. Blacklist Report: the purpose of the report is to successfully point out the enterprise customers which have been blacklisted yet have successfully been able to consume products and service of the client. This was achieved with the help of a Qlick dashboard.

Predictive model building, interpreting the results & deploying it into production as well asbuilding and maintaining dashboards for Customer Value Management Team.Built predictive models regarding the next recharge of customers for dynamic offering,customer dormancy and Churns, Dynamic offering, customer propensity based on datasources of high degree of heterogeneity and extremely imbalanced classes.Performed Social Network Analysis to detect social structures/communities on the network of various business accounts with a telecom major in middle east in order to develop a methodology for estimating market share on the network of each business accountAdHoc reports, analyses, insights or facts to test their hypothesis regarding their activity anddecision-making for Stakeholders from Sales and Marketing Teams.

Was part of the Employee Turnover Project with a Logistics Major in the US, where the objective was to identify non-union employees on the verge of attrition from the company so that a retention plan can be put in place 6 months in advance to prevent loss of talent.Specifically, the client identified two positions in the company accounting for at least 40 Percent of the attrition. Clusters were made to highlight the two most notorious job codes within all subsidiaries of the company. Further an analysis was made to distinguish between clusters with high and low attrition, and modelling was done separately for these clusters.The Data was sourced from multiple departments in the form of excel sheets. The data was joined to create new features most suitable for the project ex. 401k data, Leaves data etc. The features created were, for example, ‘Supervisor Churn Rate’ which measured the amount of attrition within the teams under each supervisor in the company.An Action item matrix was created to further assess the impact of loss of an employee and the probability to churn or attrition, and further a classification was done to identify employees at a higher or medium probability to churn with a high impact of loss.

Was part of the implementation and support of a Major FMCG based out of the Middle East. Was responsible for supporting and recreating already existing ETL pipelines in SQL. Was also responsible for initiating the monthly planning process by producing the monthly forecast across various all aggregation levels of more than 30k+ worldwide stores and 50k+ fine hygienic paper products in Oracle Demand Forecasting solution Demantra.