Sep 2020 – Sep 2025 · 5 yrs · Jersey City, New Jersey, United States · Hybrid

• 1. Engineering global strategy as the global Compliance SPOC on a project (involving complex data analyses, risk-based decisioning, both AI and Rule based components, automated integration of data vendors like BvD Orbis and Dow Jones) to implement External Entity resolution, Segmentation and Governance in collaboration with LoD1, IT, Model Validation, Data Governance and Payment Ops
• 2. Overseeing NAR FS scenarios parameter review to ensure monitoring systems are producing quality alerts with minimal false positives; designed new Threshold Tuning methodology resulting in noise reduction of ~60%
• 3. Driving key tooling projects for LoD1 and Compliance stakeholders such as roll-out of global tool to manage investigations and already delivered on SAR Reporting and Search tools and Python-based suite for seamless report generation for LoD1, LoD2, QA, SIU etc.
• 4. Strategized adoption of ISO20022 payment standards across the entire NAR AML and Sanctions program by assessing impact, proposing solution and executing delivery across Compliance processes such as Travel Rule, Subpoena search, Data Governance, Sanctions referential data screening, Scenario and Query Implementation, FINTRAC reporting; led a dedicated monthly project committee to formalize impacts and get commitments from stakeholders; presented training material, findings and solutions to multiple Group Financial Security, Compliance and Sanctions forums
• 5. Designed from scratch and executed scalable methodologies on Scenario Threshold Setting and Model Performance monitoring; all methodologies are deployed consistently across all NAR Business Lines
• 6. Ensure that procedures, methodologies and AML TM program are up to date and aligned with SR 11-7 Supervisory Guidance on MRM, DFS Part 504 on Transaction Monitoring and Filtering Program Requirements, regulatory notices and Group procedures, guidelines & policies

Mar 2019 – Sep 2019 · 6 mos

• 1. Led an NLP project on country derivation from address lines of entities across millions of payments processed by the bank every month; using n-gram frequency distribution and multi-class logistic regression in Python with governance implementation
• 2. Spearheaded Sanctions analytics - building Oracle SQL models to identify potentially suspicious vessel activity to evade US sanctions
• 3. Led the Global Markets Anti-Money Laundering (AML) project including building roadmap, acting as a liaison with the bank’s product teams and designing continuous improvement program
• 4. Led a team of 3 to implement analytics framework of NICE Actimize AML rules to detect suspicious customer behavior across US and Canada with the following components: a) replication of complex ETL and model logic using Oracle-SQL, b) optimization of model parameters to produce quality exceptions and c) risk-score calibration via user behavior and decisions to reduce operational cost by ~11 FTEs
• 5. Performed exploratory analyses to identify potential human trafficking, shell companies and terrorist-financing in the payments data leading to multiple Suspicious Activity Reports to the Financial Crimes Enforcement Network
• 6. Inculcated best practice of shared documentation and codes to better socialize domain and technical knowledge within the team
• 7. IG, Monitor and MRM Findings challenge and closure - leading closure and challenging findings and recommendations from the aforementioned teams

Apr 2016 – Feb 2019 · 2 yrs 10 mos

• 1. Remediated and enhanced ETL components based on Oracle SQL framework from data sourcing to transaction surveillance software
• 2. Designed operational efficiencies to identify false positive exceptions and volume forecasts using Oracle-SQL in partnership with Operations
• 3. Provided ancillary support to KYC by delivering VBA macros

Sep 2019 – Sep 2020 · 1 yr · New York, New York

• Led a team of data analysts on the following work-streams:
• 1. Actimize Implementation of CDD, SAM, WLF and ActOne modules
• Led all “data” tasks - acquisition, mapping, logical design and transformation, quality monitoring (DFS-504), migration and data governance
• Resolved data issues in partnership with IT and LoBs
• 2. KYC Reporting Support - Led the team of two to design Enhanced Due Diligence and Customer Transaction Geography Reports as per KYC's requirements
• 2. Enterprise Wide Risk Assessment
• Designed from scratch the data extraction & analysis approach for OFAC & BSA/AML quantitative risk assessment
• Deployed fuzzy matching algorithms to enhance customer-payment matching process
• Coordinated with KYC, Accounting and Branch IT to acquire data and Business Units to validate quantitative output (in order to demonstrate 1st LoD participation)
• Documented procedures (approach, SQL codes etc.) and approvals to make the process easily repeatable
• 3. Data Governance
• Found and covered gaps in data reconciliation process for Transaction Monitoring System in partnership with Branch IT
• Designed and evidenced data reconciliation procedure and sample reports for DFS-504 Compliance sub-certification
• As part of the BAU - reviewing and QC’ing data reconciliation reports, resolving data related issues in partnership with Branch IT and collating documentation for Internal Audit and Model Validation
• 4. AML Transaction Monitoring System (TMS)
• Led the go-live process after rule tuning project - updating existing model documentation, preparing go-live artifacts etc.
• Reviewed the current TMS components multiple times to Internal Audit’s satisfaction and provided relevant requested documents

Nov 2014 – Apr 2016 · 1 yr 5 mos · Greater New York City Area

• 1. Conceptualized and published key performance metrics with root-cause analysis and correction recommendations for management
• 2. Created word-clouds on suspicious transactions reports and used results as a feedback to improve transaction surveillance models
• 3. Presented a Tableau dashboard to the Bank Monitor in response to Dept. of Financial Services’ order to monitor customers’ address quality
• 4. Replicated existing Detection Models (implemented in Detica) in SAS; Segmentation of Primary Customers, Customer’s customer and Accounts using Clustering; Threshold Tuning of scenarios based on risk coverage; Alert Risk Scoring to prioritize alerts for investigation
• 5. Developed a macro in Excel VBA to derive country code information from Originator / Beneficiary Address Lines using text mining, ordering / sending bank country code etc.
• 6. Reviewed and corrected existing statistics in Alert / Case Burn Metrics Reporting

Jun 2014 – Nov 2014 · 5 mos

• 1. Global client based out of London, UK
• Developed logistic regression based risk-scorecard in SAS for credit card applicants with existing checking accounts
• Performed exhaustive data preparation – cleaning, missing value imputation and outlier treatment, roll ups and variable creation
• Employed Roll Rate technique to define the defaulters and Reject Inferencing techniques like Bureau Data for Rejects and Internal Performance Data to eliminate bias from the statistical models
• Used Information Value to identify significant variables, Weight of Evidence to bin variables and clustering for variable reduction
• 2. Client project: US regional bank based out of Cincinnati, Ohio
• Performed quantitative analysis via logistic regression in SAS to predict customer attrition and inactivity using survey data
• Performed text mining on ~900 million payments using sound and text matching to create modeling variables and pinpoint top competitors by accounting the outflow to clients’ external accounts
• 3. Client: One of Fortune 500 Companies, reporting process based out of New York, USA and Glasgow, UK
• Leveraged the experience of 14 sub-process owners in order to bring out the current process map and standardize all the processes
• Documented gaps and pain points in the sub-processes; listing and prioritizing the quick-fixes in order to improve the process
• Designed a standardized Excel VBA template to cater to all the reconciliation processes automation

Jun 2012 – May 2014 · 1 yr 11 mos

• 1. Business Development Project: Creating machine learning resources in R, SAS and MATLAB
• Developed company-wide training modules on machine learning techniques like non-linear regression, classification and time series modeling
• Designed code-modules and resources for techniques such as Gradient Boosting Machines, Neural Networks, Random Forests, Support Vector Machines, K-Nearest Neighbors and non-linear time series modeling using NAR, NARX and Time Delay Neural Networks
• 2. Client: One of Fortune 500 Companies based out of Kuala Lumpur and Manila
• Automated various Reconciliations resulting in multiple FTE savings; worked on various trading platforms like Sophis, Rolfe Nolan, Murex, Treats etc.

May 2011 – Jul 2011 · 2 mos · Greater Osaka Area

• 1. Project: Identifying files and applications used by employees during meetings to analyze communication and interaction phenomena
• Programmed efficient data-mining code in Python to analyze daily activity logs which were hundreds of thousands of lines long per employee
• Each employee could use multiple machines daily resulting in multiple activity logs; code filtered out the noise from the logs
• Achievement: The code was used by NEC and run on the Lab’s server in two modes, providing hourly and daily employee updates
• 2. Project: Real-time identification of files being used by employees and connection between the files
• The code extracted information on important, related and unrelated files used by the employee in real time using activity logs
• Achievement: The code was used in NEC Lab to view employees’ machine activity on the central screen