Decision Scientist

Credit Decision Scientist

CRNCY · Senior · Remote · USD contractor

Why this work matters

Across the developing world, access to credit still depends on paperwork most people can't produce. A teacher six years in the same job. A vendor who has run the same stall for a decade. Creditworthy, both — and illegible to a system built for someone else.

CRNCY exists to close that gap, using technology to open doors for the people who need them most. Every rule you write here decides who gets through.

Mission

Help CRNCY become the world's best underwriter of credit risk using unstructured and alternative data. We operate in markets where customers often have limited traditional credit information. Our objective is to make better lending decisions under uncertainty and continuously improve risk-adjusted profitability.

What You'll Do

•     Analyze portfolio performance and identify opportunities to improve profitability through credit strategy.

•     Model the probability of repayment, default and fraud from incomplete and alternative data, and build the statistical inference behind it.

•     Turn that analysis into the credit rules themselves: approval strategy, underwriting requirements, first-loan sizing and customer segmentation.

•     Optimize underwriting requirements and reduce unnecessary customer friction.

•     Design and assess experiments to improve lending outcomes.

•     Quantify trade-offs between growth, risk, customer experience and profitability.

•     Translate analysis into practical underwriting and portfolio recommendations.

You do this work yourself. There is no analytics team to delegate to. You frame the question, build the analysis, and own the decision that results from it.

What We're Looking For

Experience in at least one of: portfolio strategy or analytics, underwriting strategy, lending strategy, risk management, or another analytical role involving high-consequence decision-making under uncertainty.

Experience in consumer lending or credit is not required. We are far more interested in exceptional decision-makers than in experience in any particular industry.

We are interested in candidates who have:

•     Used data and probabilistic reasoning to make or influence important business decisions.

•     Evaluated trade-offs between competing objectives under uncertainty.

•     Applied analytical judgment rather than relying solely on predefined rules or models.

•     Translated analysis into practical business decisions and recommendations.

•     Been accountable for the business outcomes of those decisions, not simply the quality of the analysis.

Working capability, hands-on: statistical inference and probability; causal reasoning and selection bias; practical modeling — logistic regression, scorecards, gradient boosting — with clean validation discipline and no data leakage; experiment design and power; expected-value analysis; segmentation. Python and SQL at analysis-and-modeling level. You write your own code.

Required Academic Background

A degree from a strong university in a quantitative discipline: Decision Science, Operations Research, Applied Mathematics, Statistics, Economics, Engineering, or Computer Science with significant quantitative coursework.

Your training should have covered probability and statistics, decision-making under uncertainty, optimization, mathematical modeling, econometrics, operations research, or risk analysis.

Above all, we are looking for someone who naturally thinks probabilistically — comfortable making decisions with incomplete information, weighing uncertainty, and updating their judgment as new evidence becomes available.

Language: fluent English.

What This Role Is Not

This is not:

•     A Machine Learning Engineer role

•     A Data Engineering role

•     An AI Infrastructure role

•     A Machine Learning Research role

•     A traditional Data Scientist role focused primarily on model development

Candidates whose experience is primarily centered on building models, pipelines, platforms or technical infrastructure will not be considered.

What we are screening out is the candidate whose output is a model. Here, the output is a decision — you simply have to build the analysis yourself to get there.

How We Work

•     Honest and direct. Feedback here is unvarnished and immediate. You'll get it, and we expect it back.

•     Unafraid to fail. Most of what you test won't work. We would rather run the experiment and find out than defend a rule nobody has questioned.

•     Committed to excellence. High standards, applied to ourselves before anyone else.

•     Dedicated to the customer. The person on the other side of the decision is the reason the work matters.

The Setup

•     USD contractor. Fully remote, location-flexible, with meaningful daily overlap with the Americas.

•     You'll work directly with our credit and finance leadership and with the founder. Short path from analysis to decision.

•     Real ownership from day one, across multiple markets.

Requirements

The Type of Person We Need

You naturally think in probabilities, trade-offs, and expected value.

You are uncomfortable with rules that exist only because “that’s how we’ve always done it.” You instinctively ask:

• What is the probability of this outcome?
• What is the cost if it happens?
• What is the cost of preventing it?
• Is the risk worth the reward?
• What is the economically rational decision?

You are not just interested in prediction. You are interested in decision quality.

Ideal Background

The ideal candidate has worked in environments where decisions had to be made under uncertainty using incomplete or imperfect data. A degree in Decision Science, Risk Management, Economics, Statistics would be preferred.

Strong candidates may have experience with:

• Decision science, risk optimization, lending strategy, or portfolio economics.
• Customer segmentation, expected value analysis, risk-adjusted returns, or pricing optimization.
• Credit risk, underwriting analytics, scorecards, probability of default, first-payment default, expected loss, or repayment behavior analysis.
• Insurance-related risk work such as actuarial pricing, underwriting analytics, risk selection, loss forecasting, claims analytics, fraud detection, or risk-based pricing.
• Alternative data, behavioral data, unstructured data, or thin-file customer environments.
• Experimentation, causal inference, A/B testing, champion/challenger testing, Bayesian testing, or Monte Carlo simulation.
• Using messy internal data to improve real business decisions.

Technical Capabilities

This is not a pure data science research role. However, you must be technical enough to work with data, test assumptions, and answer practical modeling questions.

Helpful capabilities include:

• SQL and Python.
• Probability, statistics, segmentation, and predictive modeling.
• Logistic regression, scorecards, XGBoost, LightGBM, or similar practical models.
• Cohort analysis, vintage analysis, expected loss, customer lifetime value, and portfolio performance tracking.
• Backtesting, out-of-time validation, data leakage prevention, and scenario testing.

What This Role Is Not

This is not a general business analyst role, a pure machine learning research role, or a role for someone who needs perfect bureau data, open banking, or fully automated cash flow tools before producing useful insights.

We need someone who can work with imperfect information, think clearly about risk and reward, and help us make economically rational credit decisions.

Benefits

CRNCY offers a remote working environment, exposure to emerging-market lending, close collaboration with senior leadership, and the opportunity to help build a durable underwriting advantage using alternative data, behavioral data, internal data, and real-world outcomes.