CIFIL Research Proposal
The CIFIL Research Program drives innovation in fintech through cutting-edge studies, advanced analytics, and collaborative experimentation. It focuses on developing solutions that enhance financial security, digital infrastructure, and user experience while addressing real-world industry challenges.
With access to state-of-the-art facilities, domain experts, and interdisciplinary resources, researchers explore emerging areas such as AI, blockchain, RegTech, and data-driven finance to create high-impact, scalable, and future-ready financial technologies.
Program Structure
Research Statement 1
Research Statement 2
Research Statement 3
Research Statement 1
Bank Account Portability
Research Statement
Bank account portability refers to the ability of customers to easily transfer their bank accounts, including all associated services and transactions, from one bank to another.
Background
Every time a customer changes his/her bank for want of services he/she is required to go through the entire account opening journey again. In this process there is repetition on the part of customer to arrange for all KYC and also the track record of the existing account is not carried forward with this new account.
Research Statement 2
AI-Powered Fraud Detection and Prevention System for Banking Products
Problem Statement
Develop an advanced, AI-powered fraud detection and prevention system designed for banking products (CASA, Credit Cards, Wallets, etc.). The system should be capable of real-time transaction analysis, pattern recognition, and adaptive learning to identify and prevent fraudulent activities while minimizing false positives.
Background
In today's digital economy, financial institutions and e-commerce platforms face increasing threats from fraudulent transactions. Detecting anomalies—transactions that deviate from typical patterns—can significantly mitigate risks and protect both consumers and businesses. With the rise of big data, traditional methods may no longer suffice, necessitating innovative solutions that leverage advanced techniques in anomaly detection
Objectives:
- Data Analysis: Utilize a provided dataset of historical transactions to train and evaluate your model. Focus on identifying key features that contribute to detecting anomalies.
- Model Development: Implement at least two different anomaly detection algorithms (e.g., supervised and unsupervised) and compare their effectiveness in terms of accuracy, precision, recall, and computational efficiency.
- Real-Time Detection: Create a framework that can simulate real-time transaction processing and flag anomalies as they occur.
- Visualization: Develop an intuitive dashboard to visualize detected anomalies, providing insights into transaction patterns and flagged incidents for further investigation.
- Scalability: Ensure that your solution can be scaled to handle large datasets typically encountered in real-world applications.
Research Statement 3
Money Mule Accounts Management
Problem Statement
1. Detection and Prevention of Mule Accounts during Customer Onboarding
2. Real-Time and Near Real-Time Detection and Prevention of Mule Accounts
Background
Money mules are individuals who unknowingly or unwittingly assist criminals in laundering illegally obtained funds by receiving and transferring money through their personal bank accounts. This growing global threat has cost financial institutions millions in lost funds and reputational damage, with over $10 billion in losses reported in 2023. To combat this issue, financial institutions must implement robust measures to detect and prevent money mule accounts during the Customer Onboarding process and throughout the Customer's Lifecycle. For this hackathon competition, organized by DFS, we hereby submit the two problem statements pertaining to Mule Accounts Identification for participants to develop innovative solutions on:
Objectives
- Behavioral Analysis: Develop AI-driven systems to analyze customer details, device fingerprints, geolocation data, and document authenticity during onboarding. Focus on identifying inconsistencies, suspicious data patterns, or attempts to bypass KYC procedures.
- Transaction Monitoring: Create real-time anomaly detection systems powered by machine learning. Focus on detecting sudden bursts of activity, frequent low-value transfers, and unexplained deposits or withdrawals, enabling instant alerts and preventive actions.
- Dynamic Risk Models: Build adaptive risk-scoring mechanisms that update based on customer behavior, transactional history, and external threat intelligence. Focus on flagging high-risk accounts early and triggering additional verification steps.
- User Interaction Analysis: Leverage behavioral biometrics such as typing patterns, mouse movements, or mobile gesture analysis to detect unusual usage. Focus on building models that can differentiate between legitimate users and potential mule account operators.
- Continuous Model Training: Implement self-learning algorithms that evolve with emerging fraud trends. Focus on integrating feedback loops from confirmed cases to improve detection accuracy over time.
