Title: "Breaking the Chain: Detecting and Tracing Illicit Cryptocurrency Transactions in Pseudo-Anonymised Networks" Background: While blockchain transactions are publicly visible, cryptocurrencies like Bitcoin, Monero, and Ethereum offer a high degree of pseudonymity, making them attractive for money laundering, ransomware payments, illegal trade on darknet markets, and cross-border terror financing. Criminal actors use mixing services, privacy coins, tumblers, and exchange-hopping to obfuscate the origin and destination of funds. This poses significant challenges to law enforcement and compliance teams worldwide. Problem Statement: Design a system that can analyse cryptocurrency transactions across blockchains to detect and trace patterns indicative of illicit activity. Your solution should aim to de-anonymise suspicious clusters, identify high-risk wallet addresses, and flag laundering techniques such as mixing, chain hopping, or peel chains. Bonus points for visualising network connections or integrating with blockchain APIs (e.g., Chainalysis, BlockCypher, or Elliptic) for real-time monitoring. Key Objectives: Detect wallets or clusters involved in potentially illegal crypto activity. Identify unusual transaction patterns (e.g., layering, smurfing, rapid movement). Suggest indicators of money laundering or terror financing. Bonus: Develop dashboards or visual link analysis to aid law enforcement. Expected Output: Prototype of an AI/graph analysis-based forensic tool Data visualisations or network maps Risk scoring for addresses or transaction clusters

Title: "Spot the Fake: AI-Powered Detection of Fraudulent Websites, Apps & Digital Content" Background: As digital adoption grows, so does the threat of fraudulent online content. Fake websites mimicking banks, phishing domains, and malicious mobile apps are being used to steal data, trick users into scams, and spread malware. Many of these are sophisticated clones, and existing detection relies heavily on user reports or reactive takedowns. There's a need for proactive AI systems that can detect, classify, and flag such content before users fall prey. Problem Statement: Develop an AI/ML-based system that can automatically detect, classify, and flag fraudulent online content—including fake websites, scam mobile applications, phishing domains, or app clones. The system should leverage publicly available data (e.g., WHOIS, app metadata, DNS patterns, UI similarity) and use supervised or unsupervised models to classify content as genuine or suspicious. Key Objectives: Use NLP to detect suspicious language in websites/apps. Use computer vision to compare UI similarity with known brands. Classify apps or websites based on risk (scam, phishing, malware-laden, clone). Enable real-time alerting or browser/app warnings. Expected Output: A web or mobile prototype that scans and flags suspicious links/apps A risk classifier with explainability (e.g. “Why” it was flagged) Optional: Chrome extension or mobile app for user-facing protection