The challenge Getting an AI to work once is luck; getting it to work a million times is engineering. Any junior with an API key can write a prompt that looks like magic, but you’ll be here for the last mile—when latency spikes, models hallucinate, and the system collapses under scale. You won't be building another prototype; you are here to solve coordination problems that don't have StackOverflow answers yet, turning fragile miracles into boring, reliable infrastructure for a business that demands autonomous operations but has no map to get there.
Why this matters Deriv's mission is Trading for Anyone, Anywhere, Anytime. This scale demands AI that works in production, not prototypes that demo well. We're already here: AI resolving 65%+ of customer enquiries, writing and reviewing code, and processing invoices. We've proved what's possible—now we're scaling it. You will lead the technical implementation of our next generation of agentic systems...