Writing
I write about AI, finance, and risk,
but not in the way you'd expect.
Technical insight blended with regulatory experience, academic research with practical implementation, and occasionally, algorithmic thinking with artistic observation.
Subscribe to get new articles in your inbox.
Guides
Clear guides to AI governance, risk management, and the AIRG - start here.
Start here
AI Risk Management, from first principles
What it is, the three questions at its core, what a programme covers, and how the frameworks fit together.
The MAS AIRG for finance: the 2026 guide
The whole of the final AIRG on one page: what it requires, who it applies to, the timeline to 2027 and 2028, what changed, and how to apply it, with an FAQ.
The AIRG is final
MAS issued the final AIRG on 7 October 2026 (the consultation was November 2025), in force 7 October 2027. What changed, and what it means.
Governing AI at Scale
Keeping control of AI as the number of systems grows, in three moves.
The AIRG explained
A clear guide to Singapore's AI Risk Management Guidelines, by the person who wrote them.
AIRG vs EU AI Act, NIST & ISO 42001
How the AIRG compares to the main global AI frameworks, side by side.
AIRG, Project MindForge & the CRI Financial Services AI Risk Management Framework
How Singapore's AIRG, both MindForge documents, and the industry's control framework line up on one underlying structure.
The AIRG in practice
Inventory, risk materiality, and lifecycle controls, as a working system.
The AIRG to MindForge crosswalk
The AIRG mapped to the MindForge Operationalisation Handbook, clause by clause, with the original passages behind each mapping.
The AIRG and the FSB sound practices
The FSB's twelve international sound practices for responsible AI, mapped to the AIRG, with where they align and differ.
Governing agentic AI
Governing AI that acts: sort the branches, then cut, wire, judge, and test.
AI risk management in Singapore
An interactive timeline: FEAT, Veritas, the AIRG, the MindForge Toolkit, and more, each linked.
AI hygiene for individuals
Every AI tool is third-party. Six things to keep clean, and the red flag that catches the rest.
The five S's of using AI
Sycophancy, spirals, stagnation, slop, and spillage: the AI risks already here.
AI in Finance
Using AI in finance: generative AI and agents, investing, and forecasting, drawn from my MFE and a PhD building AI models for financial markets.
Agentic and Generative AI for Finance
How generative AI and agents work, where they help in finance, and the checks before you let them act.
AI for Investing
What AI agents can and cannot do in investing: the trust problem, the four patterns, and the gap between a demo and production.
AI for Forecasting
Why a language model cannot forecast on its own, the ladder of methods that genuinely do, and the agent that drives them.
AI Use Cases
A writeup for every use case, each broken into its chain of tasks: the data, the model, the task, and how you test it. Browse by industry, or try the interactive use-case explorer.
Industry & Policy Papers
Regulatory and industry papers I contributed to - from Singapore's AI Risk Management Guidelines to agentic-AI safeguards.
Asia Tech Alliance Co-author
Singapore Computer Society (AI Ethics & Governance Body of Knowledge) Co-author
Monetary Authority of Singapore Contributor
Monetary Authority of Singapore Lead author
Monetary Authority of Singapore Lead author
Published Research
Peer-reviewed papers on deep learning for networks, time series, and multimodal data in finance.
Lukasz Szpruch, Agus Sudjianto, Tanveer Bhatti, Gary Ang · SSRN
A. K. Saeri, J. Graham, M. Noetel, … G. Ang, et al. · arXiv
Gary Ang, Ee-Peng Lim · ACM TWEB
Gary Ang, Ee-Peng Lim · ACM TIST 2024
Gary Ang, Ee-Peng Lim · ACM TMIS 2023
Gary Ang, Ee-Peng Lim · IEEE Big Data 2022
Gary Ang, Ee-Peng Lim · ACM TiiS 2022
Gary Ang, Ee-Peng Lim · ACM TWEB
Gary Ang, Ee-Peng Lim · ACM TiiS 2022
Gary Ang, Ee-Peng Lim · ACL 2022
Gary Ang, Ee-Peng Lim · ACM IUI 2022 Honorable Mention
Gary Ang, Ee-Peng Lim · ACM ICAIF 2021
Gary Ang, Ee-Peng Lim · ACM IUI 2021
AI Governance & Risk
Pieces on AI governance, risk management, and regulation - written from the inside.
The Part That Doesn't Change
What stays constant in AI governance even as the technology and the frameworks keep shifting.
Convergence and Divergence
Why AI governance frameworks look like they are fragmenting but converge on the same underlying questions.
5 Useless Things
Things that sound essential in AI governance but turn out to be insufficient on their own.
Three Speeds of AI Adoption
Organisations adopt AI at three distinct speeds - and each requires a different governance response.
Learning risk management again, because of AI
How AI forces a rethinking of risk management from first principles.
The Darn Messy Middle of AI
The unglamorous middle ground where most AI work actually happens.
Valleys, Shifting Sands & Pillars in AI
Navigating the unstable terrain of AI governance as standards and technology shift beneath you.
A Personal Reflection on AI Risk Management
How three questions, what's at risk, how do we manage it, and who's accountable, became the spine of AI risk management.
Artificial Sweeteners or Organic Flavors? Inherent Interpretability vs. Post-Hoc Explainability
The trade-offs between building interpretable models and explaining black-box ones after the fact.
Reading Lists
Curated readings on specific AI topics - no filler, no hype.
A Simple Reading List on Third-Party AI Risk Management
Curated readings on managing AI risk across vendor and third-party relationships.
A Simple Reading List: Emergence, Not Sentience
Curated readings on emergence in AI systems - capabilities that arise without being designed.
A Simple Reading List on Data Management & AI
Curated readings on managing data for AI systems.
A Simple Reading List on Human Oversight of AI Systems
Curated readings on keeping humans meaningfully in the loop of AI decision-making.
A Simple Reading List on Explainability & Interpretability
Curated readings on making AI systems understandable to humans.
Simple Reading List on Evaluation and Testing
Curated readings on evaluating and testing AI systems.
A Simple Reading List: AI Governance & Risk Management
Curated readings on the foundations of AI governance and risk management.
Reflections & Essays
Personal writing on AI, work, art, and the spaces in between - where observation matters more than answers.
Boring Habits and AI
How boring habits from years of policy work connect to handling rogue AI agents.
About one liners, pigeonholes and niches
On picking one area, being known for one thing, and making yourself easy to remember.
Being Artfully Incompetent
Why being deliberately incompetent at some things can be worth more than trying to be good at everything.
AI and a Simple Life
Setting aside the AI FOMO to work out what a simpler life, with and without AI, looks like.
Spine, System, Unit of Life
Applying the spine, the system and the unit, the way I think about governing AI, to everyday life.
Independent Life and Portfolio Theory
Reading independent working life through the lens of portfolio theory and diversification.
Unfocused
On spending years being unfocused, and why that turned out to matter.
Worries & Independent Life
The worries that come with independent working life, and five ways to handle them.
On Authenticity
Around 300 watercolors in one video, and a question about authenticity.
My Small Answers
Small answers to big questions about AI adoption, understanding and shallow literacy.
On Happiness and Clarity
On whether independent life has made me happier, and what clarity has to do with it.
Pruning AI
What tending a bonsai teaches about pruning, shaping and living with AI.
Gradient Descent, Again
Returning to gradient descent as a way to think about navigating the uneven terrain of independent life.
The Riskiest Play
Why treating AI as the obvious next big thing may be the riskiest play right now.
The Dream and the Plan
China's small, fast, flexible approach to AI regulation, and what it suggests about dreams and plans.
Mr Kiasu vs. Mr FOMO
The final part of a conversation with Johnny Lau, on kiasu-ism, FOMO and AI.
Sum of the Parts
On combining mediocre skills into something worth more than the sum of the parts.
The Crossings
On the crossings between different worlds, and the small bubbles we each live in.
Impermanence and Drift
A photograph from 2002, impermanence, and the slow drift of things over time.
Strange Attractors Again
Making sense of a chaotic stretch of independent life through the idea of strange attractors.
Serendipity and Emergence
On rojak, serendipitous connections, and what emergence in AI has to do with either.
Thinking Out Loud
Why chain-of-thought and so-called reasoning in AI may be less than it appears.
Irony
The framework that started it all was quietly retired this week. A reflection on irony, and what comes next.
Unfinished
A meditation on incompleteness and the value of work still in progress.
Gradient Descent
An exploration of gradient descent as both optimisation algorithm and metaphor for learning.
Garfield's Eyes
Seeing the world differently through an unexpected lens.
Diffusion
How diffusion models create by learning to reverse noise.
Grounding
On the importance of grounding AI systems - and ourselves - in reality.
The Edge of Action
Exploring the boundary between thinking and doing in an AI-augmented world.
Context
Why context is the most underrated concept in AI and in life.
Alignment
On alignment - the deceptively simple idea at the heart of AI safety.
The Outsider
Reflections on working at the intersection of fields where you never fully belong to any one.
My Shifu
A tribute to mentorship and the craft of learning from a master.
Many Different Worlds
On navigating the many parallel worlds of work, identity, and AI.
The Flywheel
How momentum compounds - in business, in AI adoption, and in creative work.
Craft & Patience in the Age of AI
A case for slowness and craftsmanship when everything accelerates.
Resonance: AI at Work and in the Classroom
When AI in the workplace and AI in education share the same underlying tensions.
Phase Shifts in AI and Life
How sudden transitions in AI mirror phase shifts in physics and in personal growth.
Why I Write to Crickets (A LinkedIn Reflection for 2025)
On writing without an audience and why it still matters.
Thinking In
Long-form series on AI, risk, and networks - each building a mental model from first principles.
Thinking in...
Thinking in Attention
How attention mechanisms reshaped AI - and what they reveal about how we think.
Thinking in Uncertainty
A framework for embracing uncertainty rather than pretending it away.
Thinking in Agents: Systems thinking in the age of AI
Applying systems thinking to understand agentic AI and its emergent behaviours.
Thinking in Fundamentals
Why fundamentals matter more than ever when the tools keep changing.
Thinking in Risks: Uncertainty, Unexpectedness, and Unexplainability
The three U's of AI risk - a framework for thinking about what can go wrong.
Thinking in Networks
Thinking in Networks: A Guide to Network Tasks
A practical guide to the tasks you can solve with network-based approaches.
Thinking in Networks: Deconstructing Networks
Breaking down network structures to understand their components and power.
Thinking in Networks: The Power of Relationships
Why relationships between entities matter more than the entities themselves.
Thinking in AI
Thinking in AI 5/5: Data and Tasks in Generative AI
How generative AI redefines the relationship between data types and tasks.
Thinking in AI 4/5: Understanding data types and tasks (tasks)
A taxonomy of AI tasks and how they map to real-world problems.
Thinking in AI 3/5: Understanding data types and tasks (data types)
Understanding the data types that underpin AI - and why they shape what models can do.
Thinking in AI 2/5: Deconstructing AI: An AI Model is Just Math
Demystifying AI models by showing they are, at core, mathematical functions.
Thinking in AI 1/5: Introduction to Thinking in AI (and Generative AI)
The opening essay in a five-part series on building intuition for how AI works.
Coding & Building
Notes from actually building with AI - tools, experiments, and lessons.
OpenClaw - Kiasu Version
A distinctly Singaporean take on building with open-source AI tools.
I Built a Private AI Agent on a $12 Server. The AI Was the Easy Part.
Lessons from deploying an AI agent where infrastructure and ops matter more than the model.
Snippets: OpenAI Charges for Words, So I Sent It a Picture Instead
A playful experiment in multimodal prompt engineering to save on token costs.
Reflections: Finding the right 'vibe' with LLMs for coding
On finding a productive working relationship with LLMs as a coding partner.
If the writing resonates, the training goes deeper.
Workshops, courses, and advisory for teams who want to move beyond reading about AI to actually governing it.