@horse3903
Rafael Chong
AI & Machine Learning × Economics & Finance
CSIT Computing Scholar · Hwa Chong Institution · Singapore
CSIT Computing ScholarNOAI GoldSentinel Challenge (Open) ChampionGYEC Global Top 10
@horse3903
CSIT Computing Scholar · Hwa Chong Institution · Singapore
01
I study how intelligent adaptive agents - whether reinforcement-learning models or market participants - produce systemic outcomes no individual agent intended. The Minsky Market Sim and the CAPTCHA-LLM paper are two instances of the same question: what emerges when adaptive systems encounter structural constraints?
Trained in AI and economics at Hwa Chong as a CSIT Computing Scholar, I bring quantitative methods to problems at the boundary of technology and institution. My research spans autonomous UAV obstacle detection (NTU, under Asst Prof Mir Feroskhan), computer vision for chemistry lab error detection (SSEF Finalist, 91.1% accuracy), and a first-author arXiv preprint on LLM perceptual limits in CAPTCHA security.
Outside research: Overall Captain of Hwa Chong Track & Field (250+ members), two-time National School Games C Division Champion in the 100m and 4x100m relay, and international sprinter for Singapore.
Agent-based model
Profit-maximising RL agents endogenously generate leverage bubbles and cascading margin calls from a stable equilibrium. Features a live interactive Streamlit dashboard.
arXiv · Apr 2026
Introduced ASCII art and noise-corrupted audio as CAPTCHA classes. No frontier model (GPT-4o, Gemini) solved a single ASCII instance. arXiv preprint; peer review in progress.
JLI Economics · Jun 2026
Submitted to the John Locke Institute Economics Competition. Argues enrichment through institutional power, not mere theft.
02
Agent-based reinforcement learning simulation of Minsky's Financial Instability Hypothesis, testing whether profit-maximising RL agents endogenously generate financial fragility from a stable market equilibrium. Features a full market simulator with leverage, margin calls, and forced liquidation, Minsky finance-state classification, and an interactive Streamlit dashboard.
Investigated the viability of human-specialised neural processing as a basis for next-generation CAPTCHAs resilient to multimodal LLMs. Introduced ASCII art-based visual CAPTCHAs and noise-corrupted audio QA CAPTCHAs. No frontier model (GPT-4o, Gemini) solved a single ASCII CAPTCHA; all performed near-randomly on audio CAPTCHAs.
Research project for obstacle detection and avoidance on a DJI Tello drone using deep learning depth estimation models and image processing algorithms. Conducted under Prof Mir Feroskhan and Dr Wen Nuan.
Privacy-preserving group scheduling app that finds shared free time from Google Calendar without exposing private event details. Members see colour-coded availability overlays and can propose meeting times with real-time conflict warnings.
Chrome extension that detects scam posts on Instagram, Facebook, Twitter, and Reddit using a two-model ML pipeline - TF-IDF logistic regression on post text fused with a metadata classifier (follower counts, engagement ratios, account age). Injects a real-time risk badge directly into each post with a full analysis overlay.
03
Investigated the viability of human-specialised neural processing as a basis for next-generation CAPTCHAs resilient to multimodal LLMs. Introduced two CAPTCHA classes (ASCII art-based visual CAPTCHAs and noise-corrupted audio QA CAPTCHAs) and evaluated them against frontier models including GPT-4o and Gemini. No model solved a single ASCII-based CAPTCHA and all models performed near-randomly on audio CAPTCHAs, suggesting strong near-term robustness against current AI capabilities.
LabAssist: a computer vision system for detecting procedural errors in chemistry titration experiments. Investigated synthetic video data to improve model generalisation, comparing low-fidelity (background replacement, GMFlow optical flow) and high-fidelity (diffusion-based image animation) pipelines. GMFlow-based augmentation achieved 91.1% accuracy vs. an 83.3% real-world-only baseline, demonstrating that motion-preserving synthetic data generation is an effective strategy for scalable training dataset expansion.
04
Hwa Chong Infocomm & Robotics Society
Oxbridge Summer Scholars
Singapore Engineering and Technological Fellowship (SETF)
Amazon Web Services (AWS)
Nanyang Technological University Singapore
05
Integrated Programme (Junior College) · Promotional Exams: AAAA/B
Integrated Programme (Secondary) · MSG 1.30, 9 A1s, Higher Chinese O-Levels A2M
06
Awards, scholarships, and competition results.
AI Singapore
Highest award tier; placed among the top 12 students nationally at the pre-university level in AI concepts, applied ML, and computational problem-solving.
Science Centre Singapore
Presented original research on computer vision and synthetic data generation for chemistry lab error detection at Singapore's national science and engineering competition.
Equinor · Space Centre Houston
Recognised at the Conrad Challenge, a prestigious international STEM innovation and entrepreneurship competition supported by Equinor and Space Centre Houston.
Ministry of Education Singapore
Awarded to top-performing students in independent schools in recognition of outstanding academic achievement and conduct.
Centre for Strategic Infocomm Technologies
One of Singapore's most competitive pre-university computing scholarships, granted to students with exceptional aptitude in computing and technology.
Ministry of Education Singapore
Awarded by the Ministry of Education Singapore in recognition of strong academic performance and character at the pre-university level.
Entreplanet
Represented Singapore and placed Top 10 out of 430+ teams from 28 countries in the world's largest youth entrepreneurship competition.
NUS Entrepreneurship Society
1st place at one of Singapore's most prominent pre-university entrepreneurship competitions, pitching an original startup concept against teams from across the country.
Amazon Web Services · Delta Careers
1st place in the internal data analytics case study competition among all participants of the AWS Work Experience Programme.
Digital Intelligence Service, MINDEF Singapore
1st place nationally in Singapore's premier Capture-the-Flag cybersecurity competition, organised by the Digital Intelligence Service under MINDEF.
Singapore Mathematical Society
Presented original research on mathematical foundations of machine learning, including a proof-of-concept classification model applied to decision pattern analysis.
Interactive map of all 100+ IOAI topics across 5 tracks
53 game analyses, 11 concept chapters, 10 sessions
A self-reflection of various economics- and finance-centric studies.
