Frank Sparacino

Problem

Temporary receiving yards for large capital project modules required precise sizing to avoid over-investing in land leases while ensuring zero offloading delays or schedule risk at peak delivery.

Solution

Developed an automated Monte Carlo simulation tool in Excel using custom VBA that ingested real-time module shipping data to perform quantitative schedule and cost risk modeling across supply chain variables.

Impact

Identified the exact P80 capacity required for net peak storage, eliminating unnecessary land lease capital exposure while protecting the project critical path.

Problem

Trade contract buyouts during project kickoff often suffered from scope gaps or redundant coverage, leading to costly change orders and administrative friction across tangential trades.

Solution

Engineered a Copilot Studio AI agent utilizing Retrieval-Augmented Generation (RAG) to cross-reference current contract terms against historical trade buyouts, while scanning adjacent trade scopes to ensure 100% buyout coverage.

Impact

Standardized scope-building across trade packages, eliminated legacy scope gaps before execution, and accelerated the contract buyout cycle.

Problem

Field teams managed high-stakes commissioning milestones using static Primavera P6 PDFs, creating schedule drift, uncoordinated vendor handoffs, and hidden critical path exposure.

Solution

Engineered a custom Power App that ingested Oracle Primavera P6 schedule logic into an interactive field calendar, triggering automated vendor coordination alerts, tracking activity movement, and writing status updates back to P6 for baseline control.

Impact

Eliminated static schedule blind spots, mitigated trade stacking risks, and maintained schedule integrity during critical startup phases.

Problem

Executive recruiters, project leaders, and academic committees need immediate, context-aware answers regarding background, credentials, and technical capabilities without manual email back-and-forth.

Solution

Designed and deployed a serverless, edge-hosted AI twin using Cloudflare Workers AI and Vectorize to execute real-time retrieval-augmented generation (RAG) over structured documentation.

Impact

Demonstrates live edge architecture, low-latency RAG implementation, and custom system prompt governance directly on the web page.