Engagements and systems delivered by Regel Group’s principal for enterprise, government and private clients. Client names are withheld.
Operations research · Logistics
Explainable dispatch optimization
Designed the explanation layer for a large-scale dispatch optimization system, so planners can ask why the optimizer made a decision and get an answer traced to the underlying run data. The architecture combines a layered question taxonomy, a knowledge graph of model inputs and outputs, and graph-based retrieval.
Private AI
Offline question answering for an optimization platform
Built a self-contained question-answering system that runs with no external packages, network access or GPU. Answers are grounded in run data, a data dictionary and system documentation through hybrid lexical retrieval, a trained reranker, and a confidence gate that declines to answer rather than guess.
Statistical audit · Forecasting
Measurement program for a production demand predictor
Instrumented a rules-based demand prediction method across two facilities, capturing its output at every planning cutoff and scoring precision and recall against actual outcomes. Machine learning is gated on that baseline, so any model must beat the existing rule on evidence.
Retrieval · Human capital
Semantic candidate search across 6,000+ resumes
Built a retrieval-augmented search tool that embedded more than 6,000 resumes, letting recruiters find candidates by skills and experience instead of keyword matches.
Program management
PMO reporting and SharePoint modernization
Ran program management office reporting and led a SharePoint overhaul that consolidated project records, status reporting and leadership dashboards in one governed system.
AI literacy
AI literacy curriculum for an enterprise AI committee
Developed AI literacy training for a corporate AI subcommittee: how models work, where they fail, and how to use them safely with company data.