Staff Applied ML Engineer – Financial Crime

Overview

As Staff Applied ML Engineer in Wise’s Risk ML team, you will shape architecture strategy, ship production neural models, and scale ML for financial crime risk. You collaborate with data scientists, platform engineers, and domain experts to deliver end-to-end risk solutions that operate in real time at scale. This greenfield role offers leadership backing to push modern ML—from deep learning to foundation models—across FinCrime domains. You’ll influence direction and mentor others while delivering measurable impact.

Responsibilities

• Design and ship ML and deep learning models for financial crime detection (sequence-, graph-, and attention-based) in real time
• Define architecture strategy for applying modern ML to risk, including model families and serving/training paradigms
• Build reusable end-to-end pipelines from experimentation to production deployment
• Evaluate and prototype foundation models and embedding methods for transaction representation
• Collaborate with Data Science on model evaluation, experimental design, and causal measurement
• Mentor engineers and data scientists on ML fundamentals, production best practices, and architectural decisions

Key requirements

• Production experience shipping deep learning models at scale with real traffic and latency constraints
• Ability to make architecture-level decisions independently (model selection, training infra, serving strategy) and justify tradeoffs
• Experience designing ML systems with hard latency and throughput requirements, including optimizations like quantization and batching
• Strong fundamentals in deep learning (gradient dynamics, attention, graph message-passing, sequence modeling)
• Track record of influencing technical strategy across teams
• Proficiency in Python, PyTorch, distributed training, and ML pipeline orchestration

• Strategic influence and cross-team collaboration
• Mentoring and technical leadership
• Independent decision-making with clear communication
• Python
• PyTorch
• distributed training

…