Selected Engagements

Outcomes that defended a quarterly forecast.

Each case study below maps a real business problem to the architecture, statistical method, and reporting layer that solved it.

01 · Gaming Analytics

Pong Game Studios

+19%

Revenue Lift

GGR Optimization Engine

Redesigned slot floor performance tracking with data-driven adjustments to RTP, volatility, and hit frequency — sustaining a 19% increase in Gross Gaming Revenue and an 11% lift in player retention.

  • Lift isolated via matched pre/post windows across comparable locations, controlling for seasonality and Gen1→Gen2 migration before attributing the gain to math changes.
  • Engineered slot performance models using K-means clustering, multiple regression, and the Taguchi experimental method.
  • Directed a team of analysts in partnership with the CEO, CTO, and VP of Product.
  • Translated granular Oracle/SQL Server datasets into actionable revenue strategies through data storytelling.

02 · AI & Automation

Pong Game Studios

20hrs

Saved Weekly

AI-Agent BI Ecosystem

Architected a scalable Microsoft Fabric environment and engineered custom Model Context Protocol (MCP) servers — integrating Claude with Power BI, SQL, and Jira to reduce a full-day C-suite reporting task to seconds.

  • Savings measured against the prior manual reporting cycle (data pull, DAX, Excel assembly, narrative) — elapsed analyst hours eliminated, not a modeled estimate.
  • Built a dedicated Power BI MCP that develops semantic models from evolving business requirements.
  • Leveraged the Claude AI API to generate Python scripts executing DAX queries and synthesizing insights into Excel.
  • Eliminated roughly 20 hours of manual insight-generation work per week.

03 · BI & AI Assistants

Pong Game Studios

10–15

Requests Off-Plated Weekly

Company-Wide Analytics Portal + Dr. D

Built a live company-wide analytics portal and an in-house AI assistant (Dr. D) on top of a local DuckDB data warehouse. The portal is now the executive team's single source of truth for stakeholder reporting; the assistant handles the trivial and mid-level questions that used to reach the analytics team.

  • Stood up a local data warehouse on a VM that extracts from Power BI, SQL Server, and Oracle and consolidates everything into DuckDB format.
  • Built the portal with interactive dashboards and tables, plus a forecasting engine using Holt-Winters time-series analysis with range analysis, and a live statistical anomaly-detection system that flags issues as they occur.
  • Dr. D is a retrieval- and tool-augmented AI assistant built on an existing LLM, grounded in the warehouse's indexed company data. It answers executives' business questions directly, surfaces data-backed answers, and highlights gaps and improvement opportunities.
  • Impact: Dr. D now handles ~10–15 requests per week that previously consumed about an hour of analytics-team time each day. The portal serves as the single place for all stakeholders.
  • Built using Claude Code; live and used across the company. The public Dr. D on this portfolio is the same assistant concept, reimagined for visitors.

04 · Manufacturing

Camden Door Controls

+14%

Quarterly Revenue

Strategic BI for Operations

Spearheaded the BI strategy, leading a team to engineer data solutions that uncovered untapped market segments and lifted quarterly revenue by 14%.

  • Revenue contribution attributed to segments surfaced by the analysis, tracked against the prior-quarter baseline.
  • Designed executive dashboards that cut report turnaround from days to hours for operational leadership.
  • Bridged technical execution and business strategy with cross-functional project managers.

05 · Technology Services

Bartleby Technologies

95%

On-Time Delivery

Delivery Systems & PM

Delivered 95% of concurrent project plans on scope and on schedule across 20+ engagements, with Tableau-based tracking that shortened cycle time.

  • On-time rate measured across 20+ tracked engagements over two years.
  • Implemented Tableau dashboards to track project progress and performance metrics.
  • Shortened average delivery cycle time through disciplined prioritization and risk management.

Interactive Dashboard

The Player Segmentation Model

Six clusters, 82,608 players, K-means with RFM-extended features. Explore the live cluster map, behavioural radar, Pareto curve and per-segment CRM prescriptions.

Open dashboard →

Want the deeper version of any of these?

Request a working session