Quantitative consultant with 2.5 years of experience in AML/FCC analytics.
Exposure to end-to-end development and validation of rule-based and machine learning
models for customer risk rating, customer segmentation, transaction monitoring systems (TMS),
alert prioritization, and anomaly/fraud detection.
Experience spans multiple banking environments, including
4 months of onsite client engagement in the UAE.
Strong background in transaction-based behavioural feature engineering,
independent research, and direct client interaction.
Well-versed in regulatory documentation and
MMS / Model Risk Management (SR 11-7) frameworks.
Customer Risk Rating System:
Designed and deployed end-to-end customer risk rating models for
10,000+ customers using rule-based and ML frameworks.
Built demographic and transaction-based behavioural features,
performed variable selection, trained models
(Logistic Regression, Random Forest, XGBoost, CatBoost),
and implemented SHAP-based explainability for regulatory compliance.
Customer Segmentation:
Developed unsupervised clustering models (K-Means, GMM)
to identify transaction-behaviour-based customer segments.
Built interactive compliance dashboards using Streamlit and Plotly.
Built a production-ready segmentation system for
35,000+ customers processing
4.5M+ transactions using Gaussian Mixture Models
and custom Python packages.
Alert Prioritization (Sanctions Screening):
Developed rule-based alert prioritization using NLP-based similarity
scores and demographic features.
Model Validation:
Validated Customer Risk Rating, TMS alert prioritization,
and segmentation models covering:
conceptual soundness, data validation,
model performance (recall, precision, discriminatory power),
stability (PSI, CSI), production code review,
explainability (SHAP, LIME, PDP, counterfactuals),
and performance monitoring.
Risk Threshold Optimization:
Calibrated risk score thresholds using statistical methods
(Jenks Natural Breaks, percentile binning),
supported with visualization and regulatory documentation.
Entity Resolution / Name Screening:
Validated name screening and entity resolution systems
using NLP-based fuzzy matching techniques.
Data Validation:
Validated 5+ onboarding channels and performed
TMS ETL validation on 30M+ transactions,
identifying critical data inconsistencies.
R&D Projects
Adverse Media Screening:
Designed LLM prompts for adverse media screening,
reducing false positives by ~30%
through contextual understanding.
Anomaly Detection Dashboard:
Built an AML anomaly detection framework using
Isolation Forest and One-Class SVM with
behavioural feature engineering and
LLM-generated investigative narratives.
Graph Network-Based Fraud Detection:
Developed graph-based AML models using
network metrics and community detection
to identify money laundering networks,
supported by LLM-based explanations.
RAG-Powered CBUAE Guidelines Chatbot:
Built a compliance chatbot using
Retrieval-Augmented Generation (RAG),
vector embeddings, and Guardrails AI.
LLM-Based Investigation Application:
Developed an investigation assistant using
Gemini LLM function-calling to dynamically
extract customer, account, and transaction details.
Key Coursework:
Linear Algebra, Statistical Inference,
Regression Analysis, Optimisation,
Data Mining, Time Series Analysis,
Design of Experiments,
Generalized Linear Models,
Multivariate Analysis