About the Role
You'll take genuine end-to-end ownership of how Transak detects and prevents fraud, as part of a small, senior team. The mandate is simple to state but hard to deliver: stop fraud while approving as many legitimate transactions as possible. How you do it is yours to decide — deterministic heuristics, machine learning, AI agents, or whatever the problem demands — and the adversaries on the other side are among the most sophisticated in the world. You'll work with tens of millions of transactions of history, around 1,000 live risk signals per decision across 13 providers, and peak volumes of roughly 10,000 orders per hour. You'll also act as a data scientist for the wider business, making data accessible through the Snowflake warehouse and self-serve tooling, and help advance how the company applies AI internally.
Requirements
- Strong mathematical and statistical foundations
- A self-starter who identifies the important problem, scopes it, and acts without waiting to be directed
- Fluent in Python and SQL, able to take a model from idea to something that runs and ships
- Intellectually honest about uncertainty and rigorous in evaluating your own work
- A genuine interest in crypto, payments, fraud and the data they generate
Responsibilities
- Own fraud, chargeback and transaction-risk models end to end, from framing the question to features, rules, shipping and thresholds
- Build and tune the machine-learning and signal layer that operates alongside external risk vendors
- Set the direction of where risk modelling goes next and bring the team with you
- Act as a data scientist for the wider business, making data accessible through the Snowflake warehouse, self-serve tooling, dashboards and internal data assistants
- Take on high-leverage product, growth and experimentation problems as they arise
- Help advance how the company applies AI internally, from agentic coding assistants to internal copilots and novel uses of large language models