Machine Learning Engineer II (Underwriting ML)

Affirm; United States; Remote; USD 120,000 - 180,000; Full time

Job summary

Affirm is building underwriting ML systems that decide real-time transaction risk and expected value at checkout. You will work with ML engineers, data, and platform teams to move models from idea to production, while monitoring health as user behavior and macro conditions evolve.

You will develop underwriting models, build feature pipelines, prototype ideas, and productionize them with strong risk controls.

Qualifications

• 2+ years of experience as a machine learning engineer or a PhD in a relevant field.
• Strong Python skills and production-quality code.
• Experience with gradient-boosted trees (LightGBM/XGBoost/CatBoost) or similar.
• Experience with PyTorch.
• Experience with distributed data processing (Spark).
• Experience with ML lifecycle tooling (Kubeflow, Airflow, MLflow).
• Proficient in using AI-powered developer tools to accelerate development.

Responsibilities

• Develop and iterate on underwriting prediction models for tabular and sequential data.
• Build and scale feature pipelines and training datasets.
• Prototype modeling ideas, run offline experiments, push best approaches to production with risk controls.
• Productionize models and integrate into batch/real-time decision systems.
• Instrument and monitor model and data health; define retraining workflows.
• Collaborate with Engineering, Risk Analytics, Product, and ML Platform; communicate results clearly.

Skills

Python

Production code

GBDT models

PyTorch

Spark

Kubeflow

MLflow

Code tooling

Education

PhD in a relevant field

Tools

Kubeflow

Airflow

MLflow

Spark

Job description

Affirm is reinventing credit to make it more honest and friendly, giving consumers the flexibility to buy now and pay later without any hidden fees or compounding interest.

Affirm is reinventing credit to make it more honest and friendly, giving consumers the flexibility to buy now and pay later without any hidden fees or compounding interest.

On the Underwriting ML team, you’ll build and improve machine learning systems that make real-time transaction decisions, assessing the repayment risk and expected value of everyAffirm checkout. You’ll work closely with experienced ML engineers, platform partners, and cross-functional stakeholders to take models from idea to prototype to production, and to keep them healthy with strong measurement and monitoring as user behavior and macroeconomic conditions evolve.

What you’ll do

• You will develop and iterate on underwriting prediction models using a mix of approaches for tabular and sequential data
• You will build and scale feature pipelines and training datasets from proprietary and third-party signals, partnering with data and platform teams when needed.
• You will prototype new modeling ideas and features, run offline experiments, and drive the best-performing approaches into production with appropriate risk controls.
• You will help productionize models: integrate into batch and/or real-time decision systems, and improve reliability, latency, and operational robustness.
• You will instrument and monitor model and data health, and help define retraining/backtesting workflows
• You will collaborate across Engineering, Risk Analytics, Product, and ML Platform to define requirements, evaluate tradeoffs, and communicate results clearly to both technical and non-technical audiences.

What we look for

• You have a total of 2+ years of experience as a machine learning engineer or a PhD in a relevant field.
• Strong Python skills and experience writing production-quality code.
• Experience building and evaluating models for classification problems (preferably gradient-boosted decision trees like LightGBM/XGBoost/CatBoost, or similar).
• Experience with a deep learning framework (PyTorch preferred).
• Experience working with distributed data processing or parallel compute frameworks (Spark preferred; Ray/Dask or similar).
• Experience with ML lifecycle tooling for training orchestration, experimentation, and model monitoring (e.g., Kubeflow, Airflow, MLflow, or equivalent internal platforms).
• Proficient in using AI-powered developer tools (e.g., Claude Code, Cursor, or similar) to accelerate iteration, debugging, and code quality as part of day-to-day development workflows.
• You have mastered taking a simple problem or business scenario into a solution that interacts with multiple software components, and executing on it by writing clear, easily understood, well tested and extensible code.
• You are comfortable navigating a large code base, debugging others' code, and providing feedback to other engin