Founding Machine Learning Engineer — Risk & Fraud

About Coinflow

Coinflow is the payments infrastructure powering instant settlement for marketplaces, fintechs, remittance providers, gaming platforms, and ecommerce merchants worldwide. We combine stablecoin rails with AI-driven fraud prevention to move money faster than anyone else in the industry.

We've raised over $70M from investors including Pantera Capital, CMT Digital, Coinbase Ventures, Jump Crypto, and Reciprocal Ventures. We're profitable, we've grown revenue 23x since our seed round, and we process multi-billion-dollar annual transaction volume — making us one of the fastest-growing companies in payments.

The Role

This is a founding ML role. You'll lead our first dedicated machine learning team and own the systems that decide — in real time — which transactions we approve, which merchants we underwrite, and where fraud losses stop.

Today, risk decisions at our scale depend on a mix of internal signals and external partners. Your mandate is to build the in-house intelligence that makes those decisions better, faster, and entirely our own: models that combine our proprietary transaction data with external signals to maximize approval rates, sharpen merchant underwriting, and catch fraud before it costs us or our merchants a dollar.

You'll ship models that score live payment volume from day one. The feedback loop is immediate and the impact is measured in basis points on billions of dollars.

What You'll Do

• Design, build, and deploy real-time fraud detection and risk decisioning models across all payment methods
• Own the complete model lifecycle — experimentation, feature engineering, deployment, monitoring, and retraining
• Define and drive the metrics that matter: approval rate, detection rate, false positive rate, and chargeback rate
• Hunt emerging fraud patterns and attack vectors in high-volume transaction data before they become losses
• Build the feature store, data pipelines, and MLOps foundations the company will run on for years
• Integrate and benchmark external fraud and risk data sources against our internal signals
• Work directly with Engineering, Product, and Operations — and with the founders — to set long-term risk and ML strategy
• Hire and grow the ML team as the function scales

What We're Looking For

• 5+ years in machine learning, applied data science, or production ML roles
• Direct experience building fraud models in payments — acquiring-side experience strongly preferred
• A track record of taking models from proof-of-concept to production systems scoring live traffic
• Deep understanding of authorization fraud, card-not-present fraud, friendly fraud, and chargeback dynamics
• Strong statistical and feature engineering fundamentals on high-volume financial data
• Comfort with ambiguity — you can scope and solve open-ended problems without a playbook
• Strong cross-functional communication; you can explain model tradeoffs to engineers, ops, and executives

Nice to Have

• Background at an acquirer, ISO, PayFac, payments infrastructure company, or fraud/risk vendor (e.g., decisioning, chargeback guarantee, or trust & safety platforms)
• Experience building real-time (sub-100ms) scoring systems
• MLOps pipeline development and model monitoring at scale
• Knowledge of card network rules and chargeback workflows
• Experience as an early ML hire who built the function from scratch
• Crypto or stablecoin familiarity
  

Why This Role Is Different

Most fraud ML jobs mean tuning someone else's model inside a mature stack. Here you build the stack. Profitable company, founding-team equity, billions in live volume to train on, and a direct line from your models to the company's margins. Few ML roles anywhere offer this combination of greenfield ownership and immediate, measurable stakes.