I'm a Senior Consultant at EY Studio+ in Chicago, working where product delivery, cloud reliability, quality engineering, and applied AI meet. I've helped ship a retail platform across 100+ locations, supported a federal cloud implementation with zero audit findings, and built the observability and release practices that let enterprise teams deploy without a group chat full of prayers.
Outside client work, I design and ship open-source product systems in Swift, Rust, Python, and TypeScript. Each starts with a messy decision or workflow — finishing a document, understanding a portfolio, preparing a trip, learning a codebase — and turns it into something people can inspect, act on, and keep. I care about the unglamorous part that starts where the demo ends: failure handling, documentation, installers, updates, and evidence that the thing actually works.
Ten products, organized around the systems problems they tackle. Start with the gallery; expand a group for the full Value / Problem solved / Different story, stack, release state, and repository evidence.
| Operational & decision systems Document workflows · travel briefs · portfolio reasoning · accountable forecasts |
|||
|---|---|---|---|
 Orifold Native Mac document workflow |
 Voyalier Offline-ready trip briefs |
 FolioOrb Explainable portfolio decisions |
 Golavo Accountable football forecasts |
| Learning, research & finite experiences Codebase learning · cited research · intentionally bounded play |
||
|---|---|---|
 Codemble Playable codebase learning |
 Dusori Cited research briefs |
 Nindova Finite, private games |
| Bounded & evidence-linked systems Candidate-controlled matching · reversible contingency planning · source-linked conceptual medical learning |
||
|---|---|---|
 Nimanto Evidence-backed job matching |
 Vidha Reversible contingency rehearsal |
 PalDawn Source-linked disease journeys |
Ordering note: Projects remain ranked by total commits in each repository’s default-branch history as of August 26, 2026. This is a manual snapshot; the commit badges inside each group update live.
Operational & decision systems · 4 products
| Value | Produces finished, protected documents from mixed files in one native Mac workflow. |
| Problem solved | PDF work splinters across repair, OCR, editing, signing, and export tools. |
| Different | Keeps the entire document lifecycle local instead of stitching together specialist apps or cloud uploads. |
| Tech stack |
|
| Links |
|
| Value | Produces an offline-ready, shareable trip brief from reservations, official advice, weather, places, and plans. |
| Problem solved | Critical travel details stay fragmented across tabs, inboxes, and single-purpose apps. |
| Different | Optimizes for a reviewed departure artifact—not another itinerary builder or booking feed. |
| Tech stack |
|
| Links |
|
| Value | Turns holdings, market data, and risk context into explainable portfolio decision support. |
| Problem solved | It is hard to tell what deserves attention when holdings, risk, news, and market context are scattered. |
| Different | Keeps data and reasoning local, shows its evidence, and never connects to a brokerage or places trades. |
| Tech stack |
|
| Links |
|
| Value | Makes football forecasts accountable through pre-match predictions and a visible calibration history. |
| Problem solved | Forecast quality is hard to judge when past calls can be rewritten or selectively remembered. |
| Different | Seals forecasts before kickoff and scores the track record forward, so hindsight cannot edit the story. |
| Tech stack |
|
| Links |
|
Learning, research & finite experiences · 3 products
| Value | Turns a real codebase into a playable learning map grounded in parser-verified structure. |
| Problem solved | Unfamiliar codebases are hard to learn when tutorials and diagrams drift from the real source. |
| Different | Progress comes from proving understanding of actual files and relationships—not completing canned exercises. |
| Tech stack |
|
| Links |
|
| Value | Turns a hard question into a cited research brief with a durable evidence trail. |
| Problem solved | Search, reading, notes, and synthesis fragment across tabs and files, making claims hard to audit. |
| Different | Searches only permitted sources, preserves failures and gaps, and stores ordinary local files—AI optional. |
| Tech stack |
|
| Links |
|
| Value | Offers eight finite authored games and a separate self-ending wind-down session for private, offline play. |
| Problem solved | Many digital games are designed around endless engagement, tracking, rankings, and return loops. |
| Different | Every experience has a curtain call—no account, ads, telemetry, public leaderboard, or game that refuses to end. |
| Tech stack |
|
| Links |
|
Bounded & evidence-linked systems · 3 products
| Value | Turns candidate-confirmed evidence and allowlisted roles into reviewable job matches and application packets. |
| Problem solved | Job-search automation can invent qualifications or act before candidates approve the story. |
| Different | Uses confirmed evidence only, separates packet and action approval, and keeps external handoff disabled by default. |
| Tech stack |
|
| Links |
|
| Value | Lets people locally rehearse who should receive which documents, and when, without releasing anything; currently pre-alpha. |
| Problem solved | A contingency plan must handle prolonged silence without confusing it with proof, an emergency, or permission to release. |
| Different | A missed Check-in enters reversible Concern—not a conclusion; human verification, holds, vetoes, and Release remain planned. |
| Tech stack |
|
| Links |
|
| Value | Presents complex disease mechanisms as source-linked journeys through an interactive conceptual 3D map. |
| Problem solved | Cause and effect across multiple body systems is hard to follow when structure and explanation live separately. |
| Different | Synchronizes causal phases, system highlights, plain or clinical depth, and direct sources while explicitly bounding unreviewed synthesis. |
| Tech stack |
|
| Links |
|
The products above are the output. Agent Toolkit is a separate open-source distribution layer for the agent workflows I use to build and review them: one portable setup that installs the same curated plugin and skills through the native marketplaces in Codex and Claude Code.
| One source | Native on both sides | Explicit boundaries |
|---|---|---|
| A canonical skill payload keeps behavior aligned across clients. | Thin manifests preserve each client's own discovery, permissions, and authentication. | Setup records what it owns, stops before taking over existing plugins, and removes only receipt-managed state. |
The public preview packages tech-debt, improve-userflow-design, and council-review as one deliberately small, reviewable plugin. It copies no credentials, replaces no client configuration, and runs no background service.
| Principle | What it means in practice | |
|---|---|---|
| 🎯 | Outcome before feature | Start with the job someone needs to finish, then make every screen move that work forward. |
| 🧭 | Make reasoning inspectable | Preserve sources, calculations, provenance, and history so an answer can be challenged instead of merely trusted. |
| 🗃️ | Users own the artifact | Prefer durable formats, reversible changes, and deliberate sharing when the work or data is personal. |
| 🧪 | Quality is architecture | Design for observability, failure modes, testability, and releases that don't need a hero. |
| 📦 | Ship the whole product | Docs, installers, updates, demos, and release automation. Source code is the easy half. |
Chosen per problem, not per hype cycle.
| When | Role | Focus |
|---|---|---|
| 2025 — now | Senior Consultant · EY Studio+ | Cloud quality leadership · observability · applied AI strategy |
| 2022 — 2025 | Consultant · EY | Quality & performance engineering · federal cloud delivery |
| 2021 | IT Leadership Intern · SAP America | SAFe product delivery across Germany, the US, and India |
Education: MS, Information Systems — Kelley School of Business · BS, Informatics — Indiana University
Let's talk about decision products, learning software, workflow design, quality engineering, or turning a prototype into software people can actually use.
Portfolio & case studies