Business intelligenceCase 01
Clientlocal government
Time Series Data Explorer
OutcomeReduced time-series data access from hours to seconds for non-technical analysts
DurationSep 2025 – Oct 2025
Three practices, one toolkit. Business intelligence turns scattered data into a clear view of what is. Advanced analytics moves past reporting into prediction. Custom software is the application your team runs the work in. Each ships production-ready, documented, and built to outlast the engagement.
Dashboards, reporting layers, and the semantic models behind them. Descriptive products that turn scattered data into a clear view of what is - and what has been.
Built a self-service Power BI app that lets analysts explore and export time-series data from client's TIMELake warehouse without any SQL knowledge.
Developed a Power BI solution that automated client' monthly regulatory reporting to regulator, integrating data from multiple utility service facilities and transforming thousands of transactional records into interactive analytics.
Developed a Power BI solution that integrated financial and operational data from multiple independent regulated entity into a single analytical platform, enabling cross-hauler comparisons and data-driven rate setting for garbage and recycling collection services.
Developed a Power BI semantic model and reporting solution that automated solid waste forecast analysis, cutting manual Excel blending by 90% and enabling self-service insights for Finance Managers, Data Scientists, and Economists across client's solid waste operations.
Forecasting, segmentation, classification, ad-hoc modeling. Forward-looking products delivered as production-ready apps your team can run on a schedule.
Built an automated Python pipeline that consolidates adjacent, common-owner tax lots into single buildable sites, correcting a long-standing supply overcount in client's six-year buildable land inventory.
Built a lightweight analytics prototype to inspect regional planning simulation model calibration runs, enabling rapid loss-convergence visualization, parameter significance testing (pseudo-ANOVA/pseudo-F), and easy comparison of model specifications.
Built an automated time-series forecasting system in Microsoft Fabric that runs a multi-model competition across statistical, machine-learning, and transformer approaches to generate accurate 70-month forecasts for 39 demand series, enabling data-driven long-term capacity planning and budget development.
Built a reproducible, court-ready data pipeline that reconciled the opposing party's truck-diversion logs against scalehouse transaction records and exposed that 90% of sampled diversion claims were contradicted by the city's own data.
Full-stack web applications that replace the spreadsheets, forms, and handoffs running critical operations - internal tools, customer-facing products, AI-powered interfaces. Also where BI and advanced analytics ship as products, not just dashboards.
Designed and built a full-stack Next.js + Supabase data management platform that consolidates fragmented operational and financial data from the city Solid Waste Department into a centralized, validated database, replacing manual Excel workflows and providing the clean data foundation for a cost-of-service rate study.
Architected and developed a production-ready enterprise performance-measures database system that replaced manual Excel-based workflows for 20+ client departments, featuring role-based access control, automated validation, and real-time reporting.
Built an AI-powered tool with Claude Code that analyzes demand forecast dashboards and generates standardized narrative reports on demand, cutting manual report-writing time while enforcing consistency with client's reporting standards.