The One About Signals & Detection

One signal rarely tells the whole story. So how can AI help connect the dots? Join practitioners as they share how AI and machine learning are being used to uncover patterns across complex data and support what happens next. From detecting money mule activity through customer behaviour and exploring signals across multiple banks to evidence-grounded financial crime investigations, hear about the models, approaches, and real-world challenges behind building these systems in practice. More About the Sharings Yuge Zhang (AI Product Builder, OCBC AI Lab) will share on “Detection, Attribution, Narration: An End-to-End Pipeline for Explainable Money Mule Identification” Detecting money mule activity often means finding patterns across many weak signals rather than relying on any single suspicious transaction. Yuge will share how her team built an end-to-end pipeline combining a LightGBM classifier across 280 customer-level features, TreeSHAP attribution, and a self-hosted open-weight LLM that narrates each alert in plain language. Drawing on a live AML deployment where yield rate rose from 61% to 89%, she’ll explore why a technically correct explanation isn't always operationally useful, and how they keep the LLM communicating evidence rather than making decisions. (Technical Level: 200) Dr Renuga Kanagavelu (Senior Scientist, ASTAR) will share on “Multi-Bank Financial Crime Detection.” Financial crime can span multiple institutions, while individual banks typically see only one part of the picture. Drawing on research with Standard Chartered bank and ASTAR, Dr. Renuga will explore how signals across banks could help identify suspicious activity that may be difficult to detect within a single institution. Hear more about the research approaches being explored, alongside the technical and practical challenges of building financial crime detection systems across institutional boundaries. (Technical Level: 200) Ivan Poon (Director of AI Research & Partnerships, Standard Chartered Bank) will share on “Evidence-Grounded AI for Financial Crime Investigations.” Financial crime investigations often require analysts to piece together information across multiple sources before turning their findings into a clear investigation report. Ivan will share experimentation work on using AI to support this process while keeping outputs grounded in evidence. Explore how retrieval, task decomposition, evaluation, and human review can be combined into an AI-assisted investigation workflow, and the considerations behind using these systems to support rather than replace investigator judgement. (Technical Level: 200) More About the Speakers Yuge Zhang is an AI Product Builder at OCBC Bank’s AI Lab, where she turns AI use cases into working products for the bank’s business teams. She spent the past several years as a Data Scientist and Lead Data Scientist at the bank, building production AI systems across compliance, fraud detection, and payments, alongside broader operational workflows, drawing on 8 years of experience spanning banking, consulting, and manufacturing analytics. Yuge holds an M.Sc. in Statistics from NUS, is a certified FRM, and presented her work on explainable money mule detection at the KDD 2026 Workshop on Machine Learning in Finance. Dr Renuga Kanagavelu is a privacy-preserving AI scientist specializing in zero-trust federated learning for high-stakes FinTech environments. Her work enables secure, cross-institutional data collaboration for anti-money laundering (AML), fraud detection, and credit scoring without exposing confidential customer records. By utilizing Differential Privacy, she develops regulatory-compliant AI solutions designed specifically for global banking and financial ecosystems. Ivan Poon is a Director of AI Research & Partnership at Standard Chartered Bank where he leads applied research at the intersection of AI, knowledge systems, and complex financial workflows. His research explores how retrieval-augmented systems, ontology-grounded reasoning, graph technologies, specialist AI architectures, and human-in-the-loop evaluation can be combined to create trustworthy AI capabilities for regulated industries. His work has covered financial crime investigation, cyber risk assessment, legal review, connected-client analysis, and trade finance document examination. Ivan is particularly interested in adaptive agentic systems, semantic technologies, and bridging frontier AI research with real-world enterprise deployment. More About The Series AI Wednesdays is Lorong AI’s weekly gathering, bringing together practitioners, researchers and innovators for technical discussions on research insights, product development and engineering practices. Get involved: Learn more about Lorong AI | Speaker Sign-up | WhatsApp Community | LinkedIn | X