Senior AI/ML Engineer - Credit Risk & MLOps

Casey Walsh Senior AI/ML Engineer

Casey Walsh

Senior AI/ML Engineer

Gainesville, FL +1-239-***-**** **************@*******.*** SUMMARY

Senior AI/ML Engineer with 10+ years of experience in fintech and digital lending, specializing in credit risk modeling, real-time decision-making, and scalable ML platforms. Practical experience across consumer, SME, and secured lending products, with strong expertise building explainable, compliant, production-ready ML systems. Proven in end-to-end ML engineering, from data and feature pipelines to deployment, monitoring, and governance in highly regulated financial environments.

$10B+

annual loan originations

10M+

monthly checkout decisions

100M+

daily payment events

600+

credit risk features

TECHNICAL EXPERTISE

PROGRAMMING Python, SQL, Java, Scala/Spark, JavaScript, TypeScript, R, Bash/Shell, JSON, YAML

ML / AI Credit risk modeling (PD, LGD, EAD), supervised and unsupervised learning, feature engineering, model evaluation, hyperparameter tuning, class imbalance handling, XAI, SHAP, LIME, AutoML, ONNX

DEEP LEARNING / NLP PyTorch, TensorFlow, Keras, Hugging Face Transformers, BERT, RoBERTa, text classification, NER, PHI/PII detection, embeddings, prompt engineering, RAG, vector search

MLOPS / DATA ML pipelines, CI/CD, model monitoring, drift detection, governance, experiment tracking, A/B testing, feature stores, champion/challenger, ETL/ELT, data quality, lineage

CLOUD / PLATFORMS AWS S3, EMR, Glue, Lambda, Step Functions, SageMaker, Bedrock, Redshift, DynamoDB; GCP BigQuery, Dataflow, Vertex AI, Cloud Functions; Spark, Airflow, Kafka, Dask

APIS / ANALYTICS FastAPI, Flask, Django, REST APIs, GraphQL, Node.js, microservices, PostgreSQL, MySQL, Snowflake, BigQuery, Jupyter, Tableau, Power BI PROFESSIONAL EXPERIENCE

SoFi Senior AI/ML Engineer May 2022 - Present

• Led end-to-end development of credit risk models (PD/LGD/EAD) across loan products supporting $10B+ in annual loan originations, improving risk- adjusted approval rates by 6-9%.

• Orchestrated explainable ML models using SHAP-based attribution to meet regulatory and fair-lending requirements, reducing model review and compliance approval time by 30%+.

• Standardized feature pipelines with 600+ features across bureau data, income signals, cash-flow analytics, and behavioral indicators, improving stability and reducing retraining frequency by 25%.

• Architected ML governance and monitoring frameworks for performance, bias, and drift, decreasing post-deployment incidents by 40%+ and improving audit readiness.

• Applied LLMs and NLP to generate underwriting summaries and risk insights for analysts and operations teams, cutting manual review effort by 20- 25%.

Affirm Senior ML Engineer Jul 2019 – Apr 2022

• Led development of real-time credit approval models serving 10M+ monthly checkout decisions, improving approval lift by 7-10% at flat or lower loss rates.

• Created and optimized XGBoost and LightGBM models using 500+ behavioral, transactional, and merchant-level features, reducing early-payment default by 12-15%.

• Integrated credit risk and fraud signals into unified evaluation flows, cutting false declines by ~9% while maintaining fraud loss targets.

• Implemented model monitoring, drift detection, and automated retraining, decreasing model degradation incidents by 10%+ and improving on-call stability.

Stripe Machine Learning Engineer Sep 2016 – Jun 2019

• Designed and maintained batch and near-real-time pipelines processing 100M+ daily payment events, enabling merchant risk assessment and loan eligibility decisions for thousands of SMEs.

• Crafted 300+ merchant-level risk features from transaction velocity, refund behavior, and revenue stability, improving underwriting model recall by 15- 20%.

• Developed and productionized tree-based credit decision models used in Stripe Capital offers, reducing manual risk reviews by 40%+.

• Implemented early monitoring and retraining workflows to detect data drift and performance degradation, helping lower new-loan default rates by ~6- 8%.

Discover Financial Services Senior Data Analyst Jun 2013 – Aug 2016

• Analyzed 5M+ credit card and personal loan accounts to identify default risk, delinquency patterns, and loss drivers across customer segments.

• Engineered 200+ borrower- and transaction-level features from credit bureau data, payment history, and spend behavior to support underwriting and account-management models.

• Developed and validated logistic regression PD models that improved early-stage delinquency detection by 8-12% and informed approval and line- management decisions.

• Partnered with underwriting and risk teams to optimize thresholds and credit line policies, contributing to a 3-5% reduction in charge-off rates while maintaining growth.

EDUCATION

University of Florida, Gainesville, FL Bachelor's Degree in Data Science 2009 - 2013

Contact this candidate