Trust & Safety - LIVE Safety

- Designed and created AI/ML solutions to augment KYC processes.▪ Built automated extraction / scraping tools and pipeline to extract publicly available data of clients.▪ Integrated LLM, RAG & Embedding Models to aid in assessment of client’s risk credit score and grade.▪ Developed Vector Space Models to reinforce LLM's decisions on client's background information. ▪ Implemented Information Retrieval functions to capture insights from large unstructured data.- Conducted FX volume analysis and constructed time-series related prediction models to forecast FX refund rate volume.- Collaborated with stakeholders and relevant parties to retrieve Refinitiv data and back-test FX new pricing strategies.

The Trust & Safety division focuses on improving the platform's content safety.- Exploratory and deep-dive analysis on various products and features for issue verticals to identify anomalies or opportunities, and kick-start follow-up projects.- Text-mining analysis and NLP to identify harmful words/text within the platform.- Collaborated with cross-functional partners to enhance the detection and precision of models.- Conceptualised methods to identify and profile risky users' behaviours on the platform.- Proposed new structure, methods, and improvements for the team's internal outlier detection alert metric bot.

Business Banking Credit Administration