Director of Analytics working in healthcare procurement, clinical engineering, and operational finance. My work spans domain analysis and product engineering, including data models, import pipelines, APIs, frontend workflows, testing, deployment, and day-to-day operations. Current projects also include a football drafting game and a privacy-focused code reviewer for Codex.
LinkedIn · Email · New Jersey, USA
Production deployment · Private source · Demo by request
BiotraxIQ is a clinical engineering intelligence platform for hospital finance, procurement, and HTM teams. It brings RFP bids, medical equipment inventories, CMMS work orders, AP and PO spend, and service contracts into one governed workspace.
The RFP workflow normalizes different pricing models, finds hidden cost and scope risk, applies evidence-aware scoring, and models multi-year TCO. The inventory and equipment workflows cover device classification, EM risk, ghost assets, work-order matching, reliability, lifecycle, and capital planning. Financial and contract workflows connect spend, coverage, terms, clauses, and source-grounded redlines to reporting.
Ask BiotraxIQ provides workspace-scoped chat and MCP access across inventory, equipment, spend, contracts, and TCO. Results retain provenance, and missing evidence remains explicit instead of turning into a zero or a guessed answer.
Python · FastAPI · PostgreSQL · Next.js · TypeScript · MCP · Vercel · Render
Product overview · Open BiotraxIQ · Request a demo
Production deployment · Private source · Public product overview
AP Audit Portal is a multi-tenant accounts payable audit platform for healthcare organizations. It supports separate operator and client workflows for audit periods, credits, invoice validation, W-9 remediation, remittances, collections, and supporting documents.
Spreadsheet imports use preflight checks, reusable mappings, anomaly handling, provenance records, idempotency controls, and transactional reconciliation. Role-specific task queues are derived from the underlying audit records. Remittance matching keeps exact matches, suggestions, ambiguous rows, and unmatched work separate so an operator can see what still needs judgment.
The portal also covers collection reversals and source attribution, recovery dashboards, CFO evidence exports, signed document delivery, tenant isolation, structured audit history, and guarded release and restore workflows.
Next.js · React · TypeScript · Drizzle · Neon PostgreSQL · Cloudflare R2 · Vercel
Live product · Source private during release preparation · Public source release planned
WCDraft is a football drafting game built around tournament history from 1930 through 2026. Players assemble an all-time squad, choose a manager and formation, then run the team through a seeded group and knockout simulation.
The game includes Daily, Classic, Open Draft, Memory, and Blind Open modes. It also has accounts, saved runs, casual and ranked leaderboards, same-seed friend challenges, and versioned share links. Ranked submissions are replayed on the server before they are accepted.
The project combines a deterministic TypeScript game engine, a Python data and ratings pipeline, versioned runtime bundles, licensed public data, authentication, Postgres, social previews, and installable PWA delivery.
Next.js · React · TypeScript · Python · PostgreSQL · Vercel · Playwright
Independent fan project. Not affiliated with or endorsed by a football governing body.
Public and open source · Apache 2.0 · Release candidate
Codex Buddy Reviewer is an independent post-turn code reviewer and persistent companion for OpenAI Codex. It keeps Codex building while one or two separately configured reviewer connections inspect privacy-filtered repository changes in the background.
Buddy captures stable Git checkpoints, applies explicit privacy and authorization gates, validates and attributes structured findings, preserves useful disagreement, and reports partial failures without replacing the worker's result. It reviews and reports, but never edits, applies, or merges implementation changes.
The public release candidate supports Claude Code, Grok CLI, Ollama, and OpenCode reviewer adapters, plus five installable V2 pet companions. The native pet remains owned and rendered by Codex; Buddy owns review orchestration, progress events, receipts, and transcript output.
JavaScript · Codex plugins · Git checkpoints · Concurrent review · Privacy controls · Structured validation
Source, setup, and documentation
Production system · Private source
A healthcare facilities operations platform for project portfolios, contract renewals, follow-up risk, staleness rules, saved views, audited project updates, spreadsheet round-trips, notifications, and secure attachments.
Next.js · React · TypeScript · Drizzle · Neon PostgreSQL · Vercel
- cron-status-alert is a Hermes Agent plugin that sends a Discord message when a cron delivery fails or runs overdue.
- equipcost-forecast covers biomedical equipment lifecycle forecasting, failure curves, NPV and TCO repair-versus-replace analysis, and fleet replacement planning.
- medspend-normalize is a multi-ERP healthcare spend pipeline with vendor resolution, anomaly detection, facility comparison, lineage, an API, and a CLI.
- ghost-asset-detector uses seven explicit inventory signals to flag likely ghost assets and explain the classification.
- ortho-implant-benchmarking cross-references purchase and catalog data, ranks price opportunities, and produces reports and visualizations.
- healthcare-spend-dashboard is a Streamlit and Plotly exploration of synthetic procurement data across categories, vendors, facilities, contract mix, and PPI spend.
Recent upstream work includes statistical comparison and evaluator reliability features in EvalTrust, plus asynchronous task cleanup in RepoPrompt Community Edition.