Credit Risk Modeling
Developed a machine learning solution for consumer credit risk assessment, helping improve lending decisions through predictive analytics.
The project included feature engineering, model evaluation and probability calibration to deliver reliable risk predictions while supporting regulatory and business requirements.
The final solution provided a scalable analytical framework for consistent and data-driven credit decision making.
Live AI agent that recovers failed subscription payments. LangGraph, RAG, FastAPI, 726 automated checks · GDPR-ready · rules decide money, the model only drafts. deployed on Fly.io (http://Fly.io). Try it at paypilot.fly.dev (http://paypilot.fly.dev)
Designed an AI-powered decision support system combining quantitative models, machine learning and deep learning techniques to analyze financial markets.
The platform automated data collection, strategy evaluation and backtesting, providing a scalable framework for continuous strategy development.
The project demonstrated how advanced analytics can support systematic and data-driven investment decisions.:
Streamflow Solutions: AI Revenue-Recovery Systems
AI and automation systems that recover lost revenue: failed-payment, abandoned-cart, and reorder flows for e-commerce and fintech, deterministic by design.
Revend: AI Cart-Recovery App for Shopify
AI-driven abandoned-cart recovery for Shopify: a different message per shopper matched to their cart and hesitation, instead of one blanket discount.