Experience

Criat

• Singapore
• Financial Services
• 1 - 100 Employee

• Credit Analytics Developer

  • May 2023 - Present

  Quantitative Credit Risk Modelling - Data source replacement & credit model rebuilding. - Back testing and performance evaluation of Early Signal Warning - Existing client support Quantitative Credit Risk Modelling - Data source replacement & credit model rebuilding. - Back testing and performance evaluation of Early Signal Warning - Existing client support



Carro

• Singapore
• Motor Vehicle Manufacturing
• 200 - 300 Employee

• Data Scientist - Pricing

  • May 2022 - May 2023

  Selected Project Experience: 1. Predictive Valuation Model: - Built automated model monitoring and re-training pipeline with AWS architecture, enabling real-time pricing valuation capabilities, reducing 20+ operating hours on a weekly basis - Developed real-time AI-based pricing valuation model for Taiwan and Japan used-car markets using Python, with large size data at TB level, for the use of 40+ regional sales managers, achieving Mean Absolute Percent Error below 5.1% - Identified trends & patterns of over 1 million used car pricing & characteristic data in SEA region using Python and PostgreSQL, including data cleaning, feature engineering, and fraud detection - Responsible for overall model deployment and error check, delegating tasks to 2 junior team members 2. Loss Crystallisation Project: inventory loss valuation visualisation for C-suite management - Used AWS serverless architecture provisioned on docker container, automated modules such as SQL query, loss prediction and visualisation pipeline. Increased operational efficiency by 50 working hours per month - Performed loss prediction on 100k of inventory data, with the assumptions of 5 days, 15 days, 30days future sales - Prepared detailed PowerPoint decks and Tableau data visualizations, to provide actionable insights on business and financial strategies for c-suite level during weekly operations meetings 3. AI Pricing Report: a marketplace price distribution visualisation platform - Queried and generated marketplace used car listing histogram from different data sources using SQL and Tableau - Used distribution histogram and reference line to benchmark car valuation results with marketplace data - Displayed the latest past 5 listing records from each third-party data sources to reduce searching time - Attended weekly user debrief session, actively participated in feedback communications and transformed their requests to customized visualisation effects, improving user satisfaction Show less