Associate Data Scientist

Bureau is a unified risk decisioning platform for Compliance, Fraud, and Transaction risks. Our platform is a single decision-making engine, powered by a 1 billion+ identity knowledge graph. Over 150 Banks, fintechs, retailers, and digital platforms use Bureau to verify identities faster and stop fraud earlier globally.

Bureau has raised $50M+ from renowned Silicon Valley and global investors including Sorenson Capital and PayPal Ventures and is expanding rapidly from APAC to Americas, Europe, and beyond.

Bureau is building the infrastructure that makes digital identities and transactions safe and trustworthy for billions of people. The mission is big, the problems are complex, and the impact is real.

We hire people who want that level of responsibility. People who move fast, build systems from scratch, and care deeply about turning strategy into execution. If you want predictability or narrow scope, this won't be your place. If you want to shape how a scaling global company operates—keep reading.

• →

  Build and maintain features and components for graph-native models (GNNs, embeddings, graph algorithms) used to detect fraud, collusion, and identity risk

• →

  Support the design, training, and evaluation of statistical, AI/ML, and deep learning models for fraud and risk, under a senior data scientist's guidance

• →

  Run experiments, clean and prepare data, and do light pipeline work to help tune model performance

• →

  Write production-quality code backed by statistics and ML, and help build out ML observability for the models you work on

• →

  Document your work clearly enough that others on the team can pick it up

• BS/MS in ML, CS, Math, Stats, or a related field; 1 to 3 years in DS/AI/ML Engineering roles (internships with real shipped work count)

• Strong Python; working knowledge of SQL; exposure to a cloud platform (AWS/Databricks/Azure) is a plus

• Some experience taking a model from a notebook toward something production-adjacent, ideally with fraud, risk, payments, or compliance data — but we'll take strong fundamentals over domain experience

• Comfort with ambiguity and unlabelled data, and the initiative to ask questions and look things up rather than wait to be told

• Prior exposure to graph-based thinking (even coursework or personal projects) is a plus, not a requirement

• We hire self-motivated people and get out of their way

• We value performance, not hours worked

• Speed, ownership, and impact matter most