Building the future through Data and Mathematics

Intern, Quantitative Analyst @ PhillipCapital (IFS Capital)

Enhanced current GLM pricing model (Poisson/Gamma) to quantify rating factor relativities for private motor portfolios, exploring alternative models including Tweedie and GAM, Tweedie model achieved a Gini gain of 0.126, outperforming baseline model

Designed Monte Carlo simulations to model stochastic claim outcomes for 9,253 motor insurance policies under a Tweedie GLM framework, identifying 99th-percentile loss ratio ~4.54% above expected levels, quantifying portfolio tail risk

Developed SQL pipelines in Snowflake to compute borrower-level credit metrics (utilisation ratios, working-capital exposure, investment limits) and implemented rule-based classification to surface eligibility signals via Tableau dashboards for loan decisioning

Conducted literature review and evidentiary analysis for a multimodal ML credit risk framework spanning XGBoost, transformer-based NLP (FinBERT, LIWC), and alternative data modalities (audio, facial biometrics), informing model selection and feature engineering strategy for SME loan default prediction in a Singapore/SEA context

Intern, Strategy & Business Intelligence Analyst @ Adroit Investment Technologies

Researched on emerging AI technologies across various sectors and prepared teasers, industry outlook, statistical reports for prospective clients, achieving 33% increase in client acquisition for the Month of May

Managed components of financial statements using Power BI including Capital Accounts, Capital Calls, Capital Distribution

Operated across a full stack of a Learning Management System (LMS) for a children's learning school using Odoo platform, including UI/UX adjustments, business logic workflows, data model configurations. Design shortlisted by CEO to be used as pitching material

Implemented Excel algorithms to clean and compare 2,000 entries of investment data, enabling strategic report generation and stakeholder-ready insights

Prepared internal and external investor decks akin to M&A pitchbooks, integrating macroeconomic trends (e.g. US tariffs, air cargo rates) with company performance

Intern, Data Analytics & Management @ SCAL Academy

Managed and cleaned entries of 1,900 trainees from the Month of April to July 2024 using Excel VBA for generation of certificates

Developed an automated data entry system optimising mass certificate generation process, resulting in 75% reduction in manhours

Coordinated Ad-hoc tele calling from clients and trainees, ensuring all disputes and queries were addressed in time and reduced callback rate by 50%

Intern, Actuary @ AIA (Grinweiv)

Devised key presentations for insurance companies, namely DirectAsia and Allied World Assurance regarding Customer Acquisition and Financial Modelling, resulting in a projected 20% increase in client count in coming 1-3 years, won overall best ideation

Analysed option-based policy payoff structures (Call, Put, Straddle, Butterfly, Condor) to understand hedging mechanics in insurance product design

Appointed to be weekly Team Lead and managed a team of 6 in administrative matters, collate and report feedback regarding training and well-being to supervisor

Intern, Systems Engineer @ Defence Science and Technology Agency

Utilised visual tools to construct Command Center Dashboards for a Just-in-Time Data Switchboard used during threat and crisis management, Dashboard enhancements approved by Head of Technology for implementation

Devised a presentation of a prototype of the dashboard created, dissecting strengths and weaknesses for improvements in future development

Deep Learning for Dense Video Captioning

Implemented the Vid2Seq framework, a visual-language model that jointly localises and describes events in untrimmed video using CLIP-based visual extraction and a T5 sequence-to-sequence backbone. Responsible for implementing METEOR and CIDEr evaluation metrics from scratch and conducting ablation experiments across three CLIP encoders and T5 model sizes. Pretraining improved CIDEr from 59.62% to 70.05%, exceeding author-reported benchmarks

Wealth Wellness Hub

Full-stack wealth management platform built in response to a real Schroders problem statement, unifying traditional and digital assets into a single dashboard. Engineered a financial wellness scoring engine with LLM-powered insights via Groq's llama3-8b. Shipped scenario simulation, AI coach chat, Stripe subscription tiers, automated multilingual email reports, and full i18n across 5 languages; containerised with Docker

Project REACH Donation Dashboard

Full-stack donor engagement platform built in 4 days for a Hong Kong non-profit serving 230,000 underprivileged children, targeting a ~70% donor dropout rate. Led ideation and owned the AI-generated badge system and real-time donation leaderboard. Also shipped an automated campaign newsletter system. Built on Vue.js, Tailwind CSS, Flask, Supabase, Stripe, and Resend

Independent Equity Research: Trip.com Group (SEHK:9961)

Initiated coverage of Trip.com (SEHK: 9961) with a BUY recommendation and CNY 492 price target (+10.3% upside). Built a full financial model across income statement, balance sheet, cash flow, and DCF valuation over FY25–29E. Valuation anchored on 11.5x terminal EBITDA and 7.5% WACC; supported by Porter's Five Forces, peer multiples benchmarking, and sensitivity analysis