Team: Forensic & Litigation Consulting – Data & Analytics (EMEA)* Rules Optimization for leading UAE exchange houses – - Optimized Transaction Monitoring (TM) rules to enhance the client’s TM capabilities for regulatory compliance, post selecting 10 high-impact rules from over 70+ TM rules using a risk-based approach focused on high-risk business segments and AML typologies.- Conducted comprehensive data quality checks, thresholds outlier analysis, and historical alert reconciliation. Fine-tuned thresholds using Percentile approach and risk-based thresholds calibration, resulting in ~20% lesser alerts while maintaining robust risk coverage.- Executed TM enhancement activities focusing on improving rule logic using risk-based approaches, refining alert generation methods, alert case generation, daily rule execution strategies, limiting transactional scope, and risk coverage for historical SARs/iSTRs.* Enterprise-Wide Risk Assessment (EWRA) for UAE banks and exchange houses – - Developed an Excel-based tool for enterprise-wide financial crime risk assessment to identify money laundering, terrorist financing, and sanctions-related risk exposures to financial institution and assess the effectiveness of institutional controls to mitigate such risks. - Utilized quantitative and qualitative assessment criteria to evaluate risk exposure and residual risk across multiple risk categories.- Enhanced the EWRA tool with advanced logic, upgraded security features, additional functionalities, and better visualization.

Team: Compliance Analytics – Sanctions & Payments Screening * Enhanced Text Miner for Payments Screening – - Led team to develop a Python-based Fuzzy text matching recommendation model which creates the Sanctions whitelist for Transaction Screening tool, to reduce false positive alerts generation, by identifying the non-risky entities in the historical false positive alerts- Post extensive discussions, requirements assessment and logical thinking, the statistical techniques like Levenshtein Distance, K-means Clustering, Linear Discriminant Analysis (LDA), Natural Language Processing (NLP), rule-based business logic etc. were finalized for model- Automated the end-to-end manual whitelisting process (~100 man hours save per month) and generated ~1.5% lesser alerts

Team: Compliance Analytics – Sanctions & Payments Screening * Firco Entity Resolution Filter for Payment Screening (Firco Continuity by Fircosoft)– - Worked on rule-based alert adjudication model to reduce the false positive sanctions alerts generated by the real time Payment Screening tool, significantly saving manual alert investigation efforts. By using Text Pattern Matching on feature similarities against the Sanctions watchlist (World-Check & Geo Atlas data), its realized benefits include 22% alerts suppression, saving ~200 FTEs for US alone.- Performed Effectiveness and Efficiency testing to ensure accuracy and maintain risk coverage with lesser false positives

Team: Global Risk and Analytics – Financial Crime Risk (Anti Money Laundering - Transaction Monitoring)* Transaction Rule Engine on Google Cloud Platform (GCP) [Belgium/Netherlands/Czech Republic/Luxembourg/Switzerland] – - Pioneered a Transaction Monitoring Proof of Concept (PoC) project to deploy 16 standard customer/account tuning scenarios (sourced from Oracle MANTAS), along with segmentation, to 5 Bespoke countries to cover the risk from inadequate monitoring - Statistical procedures like Correlation, K-means clustering, Outlier treatment using IQR method etc. were explored and finalized for the model approach by critical thinking - Post multiple brain storming sessions, statistical tests like T-test, KS test, Z-test, F-test, H-Test, Levene, Bartlett, Normality test, Co-efficient of Variance etc. were shortlisted for segment merge decisions. - Guided the team to perform scenario-based threshold tuning and additional analytical activities like Data mapping, Exploratory data analysis, Critical data analysis, De minimis analysis, SAR analysis, Outlier segment analysis, Extreme thresholds analysis etc.- This lean project led to ~$1m saves in production cost (vs full fledge system), increased risk coverage & ~80% lesser unproductive alerts* Audit & IMR (Internal Management Review) [Australia/New Zealand] – - Addressed findings from Audit & IMR on data quality, reconciliation, model monitoring, correspondent banking controls, UAT, alert backlog etc. in Transaction Monitoring system (production) to meet the regulatory deadlines by following Tactical, Targeted and Strategic approach- Completed activities include proper governance & documentation, management action plans, liaison with end-to-end project teams for seamless collaboration, recurring walkthrough calls & preparing closure packs to conclude findings meeting regulatory acceptance

Team: Global Risk and Analytics – Financial Crime Compliance (FCC) [Anti Money Laundering - Transaction Monitoring]* Transaction Monitoring Post-Implementation Program [TMPIP] (Canada/Egypt) – - Led and executed the end-to-end AGILE based post detection project which focusses on validation, fine-tuning and optimizing the existing customer/account/party segments, thresholds & scenarios based on the feedbacks from investigated alerts, risk tolerance & modified scenario conditions to increase the efficiency, mitigate audit risks and meet country specific needs (User defined rules)- Played pivotal role in scripting the scenario logic in SAS and conducted threshold tuning using Brute-force and Percentile-based approach- Collaborated with cross functional teams and performed additional analytical activities like Model validation, Outlier segment analysis, Extreme thresholds analysis, Alert profiling, Above the line (ATL) & Below the line (BTL) analysis etc.- Achieved ~30% reduction in unproductive alerts and ensured comprehensive documentation for regulatory compliance

Team: Global Risk and Analytics – Financial Crime Compliance (FCC) [Anti Money Laundering - Transaction Monitoring]* Global Minimum Standard Anti-money laundering Scenarios (GMSAS) [Oman/Hong Kong HBAP/HASE/Kuwait/Argentina [On-site]] – - Being part of the team since inception, the high impact global project was developed to build and strengthen the bank’s Transaction Monitoring (TM) system to cater the regulatory needs (Deferred Prosecution Agreement- DPA 2012)- Segmentation and 26 scenarios tuning logic (sourced from Oracle MANTAS) on transactional behavior & pattern was built on customer/account/party level for improving the monitoring, anomaly detection and mitigating the money laundering risks- Statistical procedures like Correlation, K-means clustering, Outlier treatment using IQR method etc. were used in modelling using SAS- Statistical tests like T-test, KS test and other summary statistics were used for the judgement on segment merge and distributional percentile approach was used for threshold tuning and optimization, followed by model validation- Efforts led to completed DPA within regulatory deadline with proper controls, significantly reducing the unproductive alerts

Team: Compliance Analytics – Anti Money Laundering (AML)* Repair Model for Quality Control (QC) alerts [UK & HK] – - Developed Risk-based Scorecard model to enhance alert QC process quality to meet the regulatory needs. The model risk scores the alerts and creates an independent sample pool having highest propensity of error. The efforts led to increase in process quality from 19% to 55%.- Modelling approach includes Data Treatment, Dataset creation (Stratified random sampling), Fine/Coarse Classing, Binning, Dummy variables, Correlation/VIF/multi-collinearity checks (Linear Regression), Logistic regression and Score ranking the alerts for QC* Tactical Tuning on classical TM scenarios (Italy & Greece) – - Executed targeted approach to re-visit and optimize the customer/account segment thresholds for 5 classical TM scenarios on Norkom / Actimize (US) using Teradata SQL & SAS, reducing false positive alerts by ~32%.* Global Capacity Planning for Alert Backlog - - Estimated production capacity for global Investigation teams to meet dynamic alert backlog demands and developed weekly Excel VBA reports for 55+ countries.

Client: ModCloth.com - American online vintage-inspired fashion retailer (Pilot project)Objective: To help the client understand data and make data-driven decisions through multiple ad-hoc projects using SQL, Amazon Redshift & Tableau. The actionable insights resulted in a 12% increase in orders within the targeted departments and segments.* Promotional analysis – Analyze past promotions (site-wide and selected products) for optimal revenue and customer acquisition.* Open to Buy (OTB) - To assess Open to Buy (OTB) effectiveness for informed inventory decisions using tableau reports.* Other small ad-hoc projects include Be-the-Buyer program, Targeted impact assessment & trend analysis, Fraud detection dashboard etc.

Client: The Home Depot - World’s largest home improvement retailerObjective: To analyze the test experiments in retail stores using the Test & Learn Framework, which led to 24% increase in sales revenue.* Test stores selection – - To find a set of suitable test stores for conducting the experimental event, a statistical model was developed using techniques like Simple random sampling, Cook's outlier treatment, Time series analysis using Linear regression etc.- Post-production activities include enhancements to model methodology and automation of SAS scripts.* Control stores selection and Lift (Impact) measurement – - To select appropriate control stores for the test stores to conduct the experiment and accurately measure the impact, an analytical model was built using methods like Cook's outlier treatment, Z-score, Euclidean distance, Correlation, Time series analysis (Linear regression) etc.- Enhanced algorithm and collaborated with the IT team to develop a Flex UI tool with SAS backend for better control stores selection.- Prepared client-ready reports in Tableau & Excel to provide actionable insights for critical decision-making prior to nationwide event rollout.* Pre-Post methodology for Lift ($) measurement – - To accurately measure the financial impact of the experimental event in nationwide store rollout, while excluding external influences, a predictive model was built using K-means clustering, Trend analysis, and Time series analysis (Linear regression) etc.- Integrated Tableau Server to streamline execution into a seamless 1-step process, facilitating efficient reporting and decision-making.

Client: Shell Oil (UK)Objective: To support customized pricing tools for petrol pumps in Hungary, Germany, and UK resulting in ~4% sales revenue increase in 3 months- Responsible for SQL scripts that generate monthly, quarterly, and yearly sales reports on a weekly basis.- Maintained critical SQL server for minute-by-minute oil product price updates at local pumps to stay competitive.