# James Nichols — Résumé

I turn business and customer problems into working systems · Open to full-time and fractional roles

I co-founded and ran US Park Pass, owned customer rollouts at Small Improvements, and now build with AI coding agents. I connect the customer conversation, the business decision and the systems that put it into practice.

New York City metro · [james@jamesdnichols.com](mailto:james@jamesdnichols.com) · [jamesdnichols.com](https://jamesdnichols.com/) · [linkedin.com/in/jdnichols](https://www.linkedin.com/in/jdnichols)

[Download this résumé as a PDF](https://jamesdnichols.com/james-nichols-resume.pdf)

## Experience

### Independent Builder & Operator · AI-assisted software and business systems

2026 – present

Define business problems, direct AI coding agents and test the results. Work includes a voice receptionist with a verified acceptance call, margin tools, the fictional Duck Yeah demo and b.akery’s coordination alpha in development.

### Co-founder & COO · US Park Pass

2018 – 2026 · Bridgeport, CT

Co-founded and ran the storefront, fulfillment, marketing and customer support, with no developers on staff. Owned paid acquisition, lifecycle email and operating systems. We wound down after a 2026 federal policy change ended the reseller model.

- **$8M**: annual revenue at peak
- **500,000+**: lifetime orders
- **450,000**: people in the email program; contributed 25% of revenue

### Customer Success, Sales & Marketing · Small Improvements

2014 – 2018 · Berlin (remote)

Owned inbound sales, onboarding, training and rollout for customers including Pinterest, Dropbox, Reddit and Spotify. Implemented HubSpot and Aha!, trained colleagues and built a video-training program. Figures describe company growth during my tenure.

- **<100 → 650+**: paying customers over my tenure
- **<$1M → $7M**: ARR over my tenure

### Community & Marketing Management · Trion Worlds

Feb 2011 – Aug 2013

Launch community manager for Rift; started one of the first in-studio developer livestreams; ran ArcheAge's pre-launch community.

### Community & Marketing Management · Mythic Entertainment (EA / BioWare)

2007 – 2011

Ran the alpha and beta programs and convention demos for Warhammer Online, and wrote the Executive Producer's Letters.

### Community & Marketing Management · Sigil Games Online

May 2006 – May 2007

Ran the beta community and worldwide launch of Vanguard: Saga of Heroes, and stayed through the studio's close.

### Project Coordinator · 500 Group Inc.

Jan 2005 – Feb 2006

Project coordinator and executive assistant to Paolo Tiramani, on the folding home system that became Boxabl.

## Toolbox

### Commerce & acquisition

Shopify Plus · Google Ads · Meta Ads · Klaviyo · Buffer · Triple Whale · Report Pundit

### Customer, product & team systems

HubSpot · Close · Aha! · Jira · Trello · Notion

### Data & business decisions

Google Sheets · Apps Script · BigQuery · Google Cloud · QuickBooks · ShipStation · Sage Inventory Planner

### Software & AI-assisted delivery

Claude Code · Codex · Cursor · GitHub · Cloudflare Workers · Astro · Retell AI

### Current project exploration

Jev · MotherDuck · DuckDB · SQL · Python · TypeScript · MCP

Tool depth varies. I own problem definition, solution design and delivery decisions. AI agents write much of the code. I can read and modify SQL; I’m still learning to write it from scratch.

## Project: Following the margin all the way through

### The problem

Revenue-based bidding treated very different orders as equally valuable. I wanted advertising decisions to reflect what we actually kept.

### The implementation

I built a daily per-order profit feed for Google Ads, first with Sheets and Apps Script, then rebuilt on BigQuery and Google Cloud with AI assistance.

### The finding

A later reconstruction of 80,268 orders from February–December 2025 found 3,032 orders losing money after full costs. It exposed costs the earlier feed had missed.

### The limit

The rebuilt pipeline ran during wind-down. There is no clean before-and-after performance result to claim.

3,032 of 80,268 orders lost money once every cost was counted: 1 in 26.

| Cumulative cost layer | Orders losing money |
| --- | ---: |
| Revenue | 0 |
| + Product cost | 1 |
| + Fulfillment | 5 |
| + Shipping | 159 |
| + Payment fees | 545 |
| + Refunds | 3,032 |

Every order from February to December 2025, costs added top to bottom.

## Working projects

### [Duck Yeah](https://jamesdnichols.com/duck-yeah/)

A business problem you can play through. An illustrated adventure built for my MotherDuck application. Follow a decision through its consequences, inspect the records, and see how the project changed as I challenged what was useful.

I shaped the problem, story and experience. Claude Code and Codex implemented the code and queries under my direction.

The business and its outcomes are fictional. The application is a working demonstration.

### [b.akery](https://getbakery.dev/)

Helping people and AI agents keep work moving. I’m developing b.akery to carry decisions, context and corrections across people, projects and AI tools, so an operator doesn’t have to keep the whole business in their head.

The first alpha is focused on Codex and Claude. My own projects are the starting point for learning where shared context helps and where it gets in the way.

Early alpha in development. Its usefulness in other people’s work is still to be established.

### [Voice receptionist](https://jamesdnichols.com/#voice-receptionist)

From a service flow to a real call. Built a voice-AI receptionist and a consent-based callback demo. A July 2026 acceptance call verified the end-to-end experience.

I directed the integration and tested the caller experience. AI agents implemented the code.

Historical implementation example. The callback form is no longer on this site; recordings and caller details stay private.

## Looking for

### A builder. A lifelong learner. A bigger challenge.

I’m open to full-time and fractional roles where I can learn the business, own a meaningful problem and stay with it through implementation. Operations, growth, product and AI solutions are all ways into that work.

I like curious people, unfamiliar problems and the moment an idea becomes something useful. I want to help a team explore what AI makes possible, with responsibility for how it works for customers.

## Notes on the figures

US Park Pass annual revenue is at historical peak; orders are lifetime totals. Small Improvements figures describe company growth over my tenure. The 2025 margin findings are from an order-level reconstruction. The full profile is at jamesdnichols.com. This résumé is also available as Markdown at jamesdnichols.com/james-nichols-resume.md.
