Quant Finance • AI/ML • Portfolio Systems
Alex Khoo Shien How

Alex Khoo
Shien How

Building regime-aware, risk-controlled, explainable portfolio & trading systems.

View Projects
Currently working on Self-Evolving Fuzzy Ensemble (eFE) for Algorithmic Trading (FYP, 2026).

About

I am a final-year student at Nanyang Technological University (NTU) pursuing a double degree in Artificial Intelligence and Business Analytics - Recognized by the ASEAN Undergraduate Scholarship, Canada ASEAN-SEED Scholarship and Tunku Abdul Rahman University College Merit Scholarship .

My work sits at the intersection of quantitative research and applied AI — building data-driven systems for portfolio construction, algorithmic trading, risk management, and enterprise ML pipelines. I bring a product-engineering mindset to research: every model should be deployable, explainable, and robust under regime change.

Experience

Management Consultant Intern

Phillip General Insurance (Cambodia) / Phillip Capital Dec 2025
  • Spearheaded a 5-person consulting team as Team Lead, orchestrating cross-functional collaboration with C-suite stakeholders to align operational feasibility, IT infrastructure capabilities, and strategic partnership priorities on direct-digital sales strategy formulation.
  • Developed probabilistic lead-scoring and churn models using XGBoost with sigmoid and isotonic calibration to optimize customer acquisition & retention strategies, projected to generate $40M in incremental profit and 15% cost efficiency gains.
  • Engineered end-to-end machine learning pipeline including feature construction, cross-validation, and calibrated probability outputs for decision threshold optimization.
  • Applied F1-optimized decision threshold selection via precision-recall curve analysis on calibrated probability outputs to maximize the precision-recall trade-off, enabling quantitative segmentation of leads and at-risk policyholders into actionable risk-ranked intervention tiers.
  • Designed a Next.js + FastAPI full-stack real-time analytics dashboard with model retraining triggers.
View Selected Presentation Slides

Business Analyst Intern

Biobot Surgical May – Jul 2024
  • Conducted regional revenue segmentation for APAC and EMEA using 63 customer profiles, supporting pricing strategy review through analysis on client purchasing patterns.
  • Redesigned product evaluation manual in collaboration with leadership improving consistency in adoption assessment across client sites

Overseas Exchange

Singapore Management University — AY25/26 SUSEP

Local Exchange Aug – Dec 2025
  • Nominated as one of 25 students university-wide to participate in an immersive local exchange programme.
  • Courses: Entrepreneurial Finance (Venture Capital), Management Communication, IT Solution Architecture.

Fudan University, Shanghai, China — AY24/25 Summer Exchange

Summer Exchange Jun – Jul 2025
  • Courses: Chinese Financial Markets (A-), International Investment Law (A), Artificial Intelligence in Fintech (A).

Schulich Business School (York University), Canada — AY24/25 Semester 2 Exchange

Semester Exchange Jan – Mar 2025
  • Awarded Canada ASEAN-SEED Scholarship by Global Affairs Canada, Government of Canada.
  • GEM Ambassador for NTU-Schulich Business School.
  • Courses: AI in Business, Python for Finance, Strategic Management, Managerial Accounting, Prescriptive Analytics.

Selected Projects

Deep Learning LLM Quant

Quant Learning OS – Responsible AI in Education Hackathon

Multi-signal adaptive learning platform for quantitative finance preparation, fusing quiz performance, live paper trading behavioural analytics via Alpaca (fat-finger risk, revenge-trade detection, stop-loss discipline), AI mock interviews, and resume evidence into a single composite readiness score. Engineered an OpenAI function-calling agentic study planner with a 5-tool structured workflow and append-only NDJSON audit logging, a fine-tuned DistilBERT query router (3-class, 0.70 confidence threshold), a full RAG pipeline (FAISS + HuggingFace all-MiniLM-L6-v2) with LLM-as-a-Judge claim evaluation, and firm-specific interview scoring across 6 quant firms (Jane Street, Citadel, HRT, Two Sigma, D.E. Shaw, SIG) with distinct 4-dimension weighted rubrics per firm.

Multi-signal Readiness Score Agentic Study Planner DistilBERT Query Router RAG + LLM-as-a-Judge 6 Quant Firm Rubrics
Quant ML FYP 2026

Self-Evolving Fuzzy Ensemble (eFE) for Algorithmic Trading

A novel framework combining regime detection, deep temporal models (GRU/TCN + attention), concept drift detection, and RL-based fuzzy rule optimization. Backtested across multiple asset classes against benchmark indices.

Sharpe Ratio Optimization Max Drawdown Control Drift-Adaptive
Quant

Chinese Financial Markets

Analyzed Turn-of-the-Month return anomalies and Amihud illiquidity across global indices and HK-listed equities; investigated IPO spillover effects on peer-company abnormal returns and volume across China A-share, Hong Kong, and US markets using OLS market-model event studies.

Event Study Market Microstructure Cross-Market Analysis
View Repository
Quant Schulich FINE3300

Asian Option Pricing via Monte Carlo Simulation

Built a Monte Carlo pricing engine using GBM to price Asian options across fixed/floating strikes, arithmetic/geometric averaging, and call/put types for 5 contracts; conducted convergence analysis to determine optimal simulation count (N=8,500) balancing accuracy and cost; analyzed delta hedging via finite differences and sensitivity to moneyness and volatility.

Monte Carlo Simulation Option Pricing Delta Hedging
View Repository
Quant

Smart Beta Factor Diagnostic Framework

Engineered a 9-factor smart beta diagnostic framework in Python, running full-sample Newey-West HAC regressions and rolling 36-month stability checks across 11 ETFs to decompose factor exposures (MKT, SMB, HML, MOM, BAB, QMJ, RMW, Carry, Illiquidity) and surface alpha, IR, and PASS/WARN/FAIL verdicts for portfolio due diligence. Constructed live alternative factor series from scratch — G10 FX carry (rank-weighted long-short via FRED rates + yfinance FX) and Amihud (2002) IML (30-stock panel, prior-month sort to eliminate look-ahead bias).

Newey-West HAC Rolling 36M Stability 9 Factors • 11 ETFs PASS/WARN/FAIL Scoring
View Repository
ML Computer Vision

Kaggle Cassava Leaf Disease Classification

Hybrid ensemble approach combining ResNeXt, ViT, EfficientNet, and CropNet architectures with stacking and voting strategies for robust plant disease detection.

90.1% Accuracy Ensemble Stacking
View Repository
NLP ML

NLP Sentiment Classification

Developed and fine-tuned a BiLSTM model for sentiment classification using Optuna hyperparameter optimisation, incorporating multi-head self-attention and min-max pooling frameworks to achieve 79% accuracy.

79% Accuracy BiLSTM + Optuna Multi-head Self-Attention Min-Max Pooling
ML Computer Vision Robotics

Autonomous Maze Robot – YOLOv8 Image Recognition

Developed and fine-tuned a YOLOv8 object detection model in Python as part of a robotic system capable of autonomously traversing a maze while detecting images displayed within the arena. Automated image annotation and augmentation pipelines to generate and label 35k+ training images, significantly improving model training workflow efficiency.

85.7% Accuracy 35k+ Training Images Real-time Inference Automated Annotation
ML RL

Reinforcement Learning Double Q-Learning

Developed a Reinforcement Learning agent using Double Q-Learning with Prioritised Experience Replay and Epsilon-Greedy Policy to balance a pole on a cart under constraints, achieving an average reward exceeding 195 within 100 episodes.

Avg Reward >195 100 Episodes Prioritised Experience Replay Epsilon-Greedy Policy
View Repository
Databases

Electric Vehicle Conversion Proposal

Conducted an end-to-end feasibility study on transitioning Singapore's entire vehicle population to electric vehicles, benchmarking against the Singapore Green Plan targets. Built Tableau dashboards to visualise adoption trends, infrastructure gaps, and policy alignment. Leveraged SQL and MongoDB to query and analyse EV registration data and charging station geographic distribution across the United States as a comparative case study.

SQL & MongoDB Tableau Feasibility Study Singapore Green Plan
Databases LLM

LLM-based SQL Plan Explanation & Cost Comparison

Utilised a Llama3-8B-8192 model with text parsing techniques to generate natural language explanations of PostgreSQL query execution plans and perform automated cost comparison between the default SQL Plan and the Alternative Query Plan (AQP).

Llama3-8B-8192 PostgreSQL SQL vs AQP Cost Analysis
View Repository
LLM Agentic RAG

Agentic RAG Insurance Retention System

Built an Agentic RAG-based insurance churn prediction system using LangChain, LangGraph, and Groq (Llama 3.3 70B), integrating FAISS vector search over policy documents and client notes with custom tools for churn scoring, policy guardrail checks, and automated CSV reporting — enabling grounded, auditable retention action plans for at-risk clients.

LangChain & LangGraph Llama 3.3 70B FAISS Vector Search Automated CSV Reporting
LLM RAG

Transformer-Based Resume Evaluation & RAG Pipeline

Implemented robust multi-format document parsing and ingestion (PDF, DOCX, TXT), transforming unstructured resumes into clean, machine-readable text for downstream transformer-based embedding, retrieval, and LLM analysis. Built a transformer-based RAG system combining Sentence-Transformer semantic retrieval (FAISS) with LLM-constrained JSON extraction to enforce evidence-backed outputs and generate interpretable career-match scores.

Sentence-Transformers FAISS Retrieval RAG Pipeline JSON-constrained Extraction
View Repository
ML Healthcare

Parkinson Disease Prediction

Predictive classification model for early Parkinson's disease diagnosis using biomedical voice measurements and machine learning techniques.

Classification Biomedical Data
View Repository
ML Healthcare

Heart Disease Predictive Diagnosis

Machine learning pipeline for predictive diagnosis of heart disease using clinical and demographic patient data with multiple classification algorithms.

Classification Clinical Data
View Repository

Research Interests

Quantitative Finance

  • Market regime detection & classification
  • Portfolio construction & optimization
  • Risk overlays & drawdown management
  • Drift monitoring & model explainability

AI / Machine Learning

  • Sequence models & temporal architectures
  • Probabilistic calibration
  • Ensemble learning & model stacking
  • MLOps & production ML systems

Skills

Languages

Python R C/C++ Java SQL MongoDB

ML / DL

PyTorch TensorFlow NumPy Pandas scikit-learn XGBoost

Tools & Platforms

Bloomberg Terminal AWS Tableau Power BI CapitalIQ Git

Soft Skills

Intercultural Communication

Certifications & Activities

Certifications

  • London Institute of Banking and Finance — Level 4 Diploma of Applied Finance 2024
  • Bloomberg Market Concepts (BMC) 2024
  • BlackRock Sustainable Investing Across Asset Classes 2024

Activities

NTU Investment Banking Club — Asset Management Analyst

  • Advocated a buy position for eBay by analyzing its strategic positioning for sustainable growth, aligning with key trends and prudent capital management using different valuation methods (DCF, relative valuation).
  • Pitched a sell position for Nintendo with theses on cyclical nature of earnings, intensified market competition within the gaming industry, and global supply chain disruptions resulting in hardware procurement difficulties.
  • Advocated a buy position for Ralph Lauren with theses on expanding operating margins via a strategic pivot towards a global direct-to-consumer model and strong brand positioning.

Awards & Achievements

2nd Global

Southeast Asian Mathematical Olympiad X (Champions Division)

2022

Champion

University Science Malaysia Maths Quiz (Team)

2017, 2019

Gold Team

Singapore and Asian Schools Math Olympiad

2016

Silver & 2nd National

Southeast Asian Mathematical Olympiad

2021

Silver

Singapore and Asian Schools Math Olympiad

2016 – 2019, 2021

Silver

IMONST International Mathematical Olympiad Malaysia

2020