Large Language Models & Non-Fungible Tokens

• Developed an internal automated image forensic tool to detect manipulated images, incorporating trained Convolutional Neural Networks on ELA and PCA outputs, achieving 91% accuracy and an F1 score of 0.9, cutting down investigation time by 50%• Created agentic Retrieval-Augmented Generation LLM workflow proof of concept using Llama3, Langchain, Langraph, LlamaIndex to decrease hallucinations and improve context accuracy • Wrote a research paper on the best practices for combating AI-driven fraud, including C2PA and CAI implementation, IMDA AI Verify governance testing

- Led weekly risk assessments and performed in-depth analysis on auction ads across global markets, successfully identifying and addressing risk violations and emerging trends.- Kickstarted the launch of the SEA Accelerator Program, empowering agencies to effectively manage bad actors through the implementation of blocklisting and suspension mechanisms utilizing cutting-edge technologies like JEDI (GNNs) and Shark (RegEx).- Spearheaded the development of visualization dashboards to enhance reporting efficiency, support cross-functional teams, and enable real-time monitoring of key risk metrics.- Conducted post-opening of Forex market assessment, combining a risk assessment matrix and detailed case deep-dives, to thoroughly evaluate the impact of the Forex market on SEA markets. Delivered valuable insights and recommendations regarding the safety and viability of Forex within these markets.- Created audio to text translator using OpenAI speech recognition model

- Adviced the commercial team in category trends and distribution gaps; drove the creation of an interactive dashboard to help with insights and analysis for category market share, basket penetration, customer penetration and product distribution dashboard(ND, PCV, ACV, SPPD)- Adviced the commercial team and pandamart suppliers in optimizing promotions via a Market Basket Analysis using Pyspark framework (FP-growth algorithm) to identify and provide key insights to customer buying patterns across all categories- Analyzed monthly promotions and designed a forecasting framework to help the commercial team optimize promotion strategy and supplier funding- Team leader for Foodpanda's hackathon challenge (Overall champion). Proposed algorithm that segments sales of goods into selling velocity to then be further segregated to maximise picker and prep time efficiency- Led recurring consultative sessions to help address business issues via dashboards using Google Datastudio & BigQuery- Performed analysis for campaigns/promotions/experiments to understand better on promotion mechanics and sales performance

- Outstanding Serviceman of the Month (March)- Went for Guam Detachment from May 2021 to July 2021- Worked with NCS as a software tester during the RSAF digitalization rollout phase, testing out bugs and providing feedback to increase software capabilities.- Developed Munition logistics & Jet related component tracking software for RSAF F16CD fleet