Hong Kong · Data, AI & Quantitative Research

Hi, I'm Aylon 👋

Data Scientist· Quantitative Analyst· Applied AI Engineer

I build reproducible multi-asset research across signals, portfolio construction, risk and attribution — with applied AI where it makes the research process more rigorous.

Multi-asset allocation Signal validation Auditable AI systems

View selected work Research archive

LinkedIn GitHub Research notes

ABOUTResearch profile

I work across data science, applied AI and quantitative research. I build reproducible data and modelling workflows, design agentic systems with evaluation and auditability, and test investment ideas from signal validation through portfolio construction, costs and attribution.

Data Science Applied AI Quantitative Research Research Engineering

01 / EXPERIENCECurrent and research roles

Experience.

My research and professional journey.

2026 — PRESENT

Modelling & Data Scientist

Insurance Group

Build reproducible data pipelines, model evaluation workflows and auditable AI systems for financial processes.

Tech stack

Python

pandas

NumPy

Jupyter

scikit-learn

PyTorch LLM APIs

Git

Capabilities

Data pipelines Model evaluation AI systems

2024 — 2026

Research / Teaching Assistant

The Hong Kong Polytechnic University · Faculty of Business

Supported empirical research, data preparation and teaching delivery across quantitative and business analytics work.

Tech stack

Python

pandas

NumPy

scikit-learn

Jupyter

LaTeX

Git

Capabilities

Empirical research Data preparation Teaching

2022 — 2024

Project Researcher

Imperial College London · Department of Physics

Contributed to applied research involving quantitative analysis, computational methods and evidence-led problem solving.

Tech stack

Python

NumPy

pandas

PyTorch

Jupyter

LaTeX

Capabilities

Quantitative analysis Computational modelling Applied research

02 / EDUCATIONAcademic foundation

Education.

Formal training in mathematics, analytical reasoning and applied technical work.

Imperial College London

MSc Security and Resilience

10/2022 — 06/2024London, United Kingdom

Research direction: Risk assessment · Cyber security · Systems resilience

University of Reading

BSc Mathematics and Applied Mathematics

09/2018 — 06/2022Reading, United Kingdom

Research direction: Probability · Mathematical statistics · Stochastic processes · Optimisation

03 / BLOGResearch & practical guides

Research notes, allocation and practical guides.

Public work on multi-asset allocation, portfolio implementation and reproducible research — plus a practical data-science workflow.

01 / FEATUREDAsset allocation

Asset allocation is not one answer.

A research map from benchmarks and risk allocation to implementation and governance: 60/40, Permanent Portfolio, All Weather, risk parity, 1/N, rebalancing, constraints, liability-driven allocation and L1.

SAA / TAA Risk allocation Portfolio construction Attribution

↗

02 / PORTFOLIOThe baseline

The 60/40, tested

The most famous allocation rule breaks when stocks and bonds fall together. A 40–50% equity weight had a better drawdown in the sample.

60/40CorrelationDrawdown

↗

03 / PORTFOLIORisk balance

All Weather, tested

Steadier than a 60/40, but not environment-neutral: long bonds carried more than half of the model risk.

RegimesDurationRisk contribution

↗

04 / PORTFOLIOEqual risk

Risk parity, tested

Equalising risk makes the risk distribution cleaner, but the return and leverage trade-offs remain visible.

Risk budgetLeverageSharpe

↗

05 / CASE STUDYFrom view to weight

L1: the benchmark won. The process remains.

An end-to-end multi-asset shadow portfolio: SAA, a trend overlay, costs, risk contribution, return attribution and frozen decision rules.

8 ETFs221 monthsNegative result retained

↗

Practical guide Workflow · documentation · reproducibility

06 / PRACTICAL GUIDEData science workflow

Data Science Workflow · CRISP-DM Tutorial

A hands-on guide that turns a business question into a reproducible, evaluated data-science decision through the six CRISP-DM phases and three case labs.

Business understandingData preparationEvaluationDeployment

↗