Interview: a Singapore data science lead

Describe your day-to-day as a data science lead integrating machine learning into volatility analytics.

Most days start with a mix of code review and meetings with analysts running live volatility models. Integrating machine learning isn’t just about writing new algorithms—it’s about questioning if the data actually supports the prediction. We regularly host technical review sessions where every line of reasoning is up for debate. It’s a culture of critical feedback, not hierarchy.

What are the highlights and main challenges in your role?

The most rewarding moments come when a collaborative review surfaces an issue no one saw alone. The challenge is adapting models fast enough to keep pace with new market data. Technical debt can build quickly if you’re not vigilant. Our team works on rapid prototyping, but we have to balance speed with careful validation.

How has machine learning changed the volatility analytics process for your team?

Machine learning has given us flexibility and the ability to spot relationships missed by traditional models, but it also adds a layer of abstraction. Now, explainability is a key requirement. We spend as much time on model interpretability as we do on performance. Regular knowledge sharing is built into our process, so nobody operates in isolation.

What advice would you give to those starting out in volatility analytics?

For new analysts: prioritize learning from peers, not just textbooks. Peer review exposes blind spots you can’t see yourself. Don’t be discouraged by a steep learning curve—everyone goes through it. Document everything. That’s what will save you in a crisis.

How has your approach to data science in volatility analytics evolved over time?

At the start of my career, I thought technical skill was enough. Now, I know that being open to feedback and sharing both wins and mistakes is what drives long-term progress. No one gets it right the first time. The field rewards those who adapt quickly and share openly.