You're leaving the safety of Jupyter notebooks to wrestle with the tension of stochastic models in a deterministic financial world. You'll encounter a codebase processing millions of transactions where it works on my machine is not a valid defence, bridging the gap between tutorial implementations and high-scale systems to build co-intelligence that decides and executes without a human holding its hand.
See what we're shipping at Deriv.
Why this matters
Deriv's mission is Trading for Anyone, Anywhere, Anytime. Millions of traders, around the clock. 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, processing invoices, and screening candidates. Not experiments. Production systems you'll help extend.
Why Deriv
Learn by building, not by watching. Here's where you'll do it:
Customer experience: Building AI that handles conversation, outreach, and lifecycle management.
Developer infrastructure: Building the systems that build systems (Spec-to-PR, QA automation, security scanning).
Business functions: Building the AI that runs Deriv (finance workflows, HR automation).
Your placement depends on team needs and your interests. You'll likely focus on one area but touch several, with real ownership and support along the way.
What you'll do
Build features that go to production: You won't just write scripts; you'll ship code that runs in live environments.
Work across three paradigms: You'll learn to combine deterministic systems (code), predictive models (ML), and agentic systems (LLMs).
Learn from failure: You'll understand why guardrails matter when a 1% error rate means thousands of wrong decisions.
Pair with experienced engineers: You'll own small features end-to-end with guidance from senior mentors.
Who you are
You write code that runs: Python or another language you genuinely enjoy. You know syntax is easy; making things work in production is where it gets interesting.
You've touched ML or LLMs: Courses, side projects, experiments. Enough to know what you don't know yet.
You deliver reliably: You distinguish urgent from important and keep your promises on delivery dates.
You're comfortable being wrong: You'll ship code that breaks. That's how you learnif you can admit it and fix it.
Tech stack
Languages: Python, TypeScript
AI/ML: OpenAI APIs, Anthropic APIs, LangGraph, Custom ML Pipelines
This is demanding work. You'll face problems without clear answers. You'll ship code that breaks and fix it under pressure. Some weeks will be frustrating.
But you'll ship AI that runsnot demos, not prototypes. You'll see your work handling real transactions. And you'll grow fast.
