Analysis of the effect of different types of financial products on financial wellbeing.
-
Updated
Oct 7, 2020 - Jupyter Notebook
Analysis of the effect of different types of financial products on financial wellbeing.
HMDA mortgage lending disparity analyzer — denial rates, racial disparities, lending deserts, and lender benchmarking
Analytics pipeline that turns CFPB consumer complaints into operational intelligence: topic discovery, interpretable risk scoring, and executive reporting.
A copy of the Consumer Complaints Database from the U.S. Consumer Financial Protection Bureau downloaded on April 25, 2018.
An end-to-end analytics platform designed to process and analyze real consumer complaint data from the CFPB. By integrating demographic data and applying NLP, the platform identifies high-risk financial products and response efficiency gaps. An interactive dashboard enables real-time monitoring of key compliance metrics.
Benchmarking consumer complaints for 10 major U.S. banks on the CFPB database (16.9M rows), built end-to-end on Snowflake → dbt → Tableau with size-neutral comparisons.
Personal website (outdated)
Truist digital banking competitive benchmark: 598K CFPB complaints + 2M app store ratings across BofA, Chase, Wells Fargo, PNC, Fifth Third — with RICE-scored roadmap
R analysis of financial conditions, financial well-being, and reported distress using the CFPB National Financial Well-Being Survey.
PropTech-specific profile of the AI Incident Card. Maps severity/type fields onto CFPB UDAAP, ECOA Reg B (12 CFR 1002), Fair Housing Act, RESPA Section 8, title-chain integrity (ALTA), state real-estate regulator notification + a 5-grade transaction-impact severity scale. PropTech-readiness scaffolding, not certification.
Open-source PII detection and re-identification risk benchmarks for LLM pipelines under EU AI Act, HIPAA Safe Harbor and GLBA NPI.
Profile of evidence-bundle-spec scoped to CFPB + OCC + FRB + FDIC readiness across 8 obligation families: model-risk-management + ECOA Reg B + FCRA Reg V + GLBA Safeguards + BSA/AML + Section 1071 + Section 1033 + CFPB UDAAP. FinTech scaffolding, not certification.
Automatic Tagging of CFPB Complaints using LLM's and Ensemble Techniques
Fine-tuned DistilBERT for CFPB consumer-complaint routing (8 classes, macro-F1 0.85), plus a quantitative faithfulness audit of its explainability layer: Integrated Gradients, SHAP and attention rollout agree on only 13-29% of their top-attributed tokens.
FinTech audit-stream Operator: per-consumer-credit / deposit / payment / fraud / AML / robo-advisor / Section-1071-small-business AI tool events hash-chained for CFPB + OCC + FRB + FDIC + ECOA + FCRA + GLBA + BSA-AML evidence. Two invariants: human-credit-officer + FCRA permissible-purpose. Excludes mortgage + insurance.
Production-grade RAG system over CFPB consumer financial complaint narratives featuring FastAPI REST API, Streamlit UI, Chroma vector store, anti-hallucination guardrails, and Docker deployment.
n8n workflow orchestrating four conditional-tool-use Claude agents to classify, research, draft, and QA real CFPB consumer complaints against real federal regulation text, behind a deterministic escalation gate. Live Streamlit dashboard synced to Google Sheets. Built to prioritize real, disclosed data over simulation throughout.
To associate your repository with the cfpb topic, visit your repo's landing page and select "manage topics."