I use AI and automation to build tools that solve real problems — from intelligent data pipelines to LLM-powered business workflows.
An end-to-end intake system that ingests stakeholder analytical requests, classifies them by data product area and priority using the Claude API, and auto-generates structured requirement specifications — directly mirroring enterprise data demand register management. Non-technical stakeholders can submit requests in plain English and receive structured specs back automatically.
Cohort retention analysis on 50K+ records to identify churn patterns. Pinpointed a critical 60-day drop-off responsible for 35% of monthly churn. Applied logistic regression model achieving 84% prediction accuracy with Tableau dashboard for leadership review.
End-to-end ETL pipeline with automated reconciliation logic integrated into financial workflows. Anomaly detection flags exceptions before reporting — reducing manual consolidation by 45% and exception resolution time by 35%.
An AI-powered document assistant built with Claude API that answers natural language questions over uploaded business reports, financial statements, and SOP documents. Drop a PDF, ask a question, get an answer with source reference.
A conversational BI interface where business users can ask "Why did revenue drop in Q3?" and receive an auto-generated analysis with charts and root-cause explanations — no SQL required. Powered by Claude + SQL generation pipeline.
Time-series demand and revenue forecasting tool using ARIMA and Prophet models. Provides confidence intervals, seasonality decomposition, and scenario modeling — outputs formatted as Excel-ready reports for finance teams.