It can only propose
Nothing changes on the board until a person clicks Approve. Not a setting, not a preference somebody can switch off in a hurry. The tool has no way to move your work on its own.
Staff Design Operations Program Manager · New York
A ticket queue tracks the work.
A ticket queue tracks the work.
An operating system moves it.1
Most Design Ops is a service desk: a queue, a calendar, a tool inventory. Your posting asks for the other thing, the function that makes a design org faster the bigger it gets. I have built that function from nothing once and hired the team into it. I am the operating layer of a brand function now. And when the process needed a tool that did not exist, I built that too, with AI, and it runs every week.
Diligent on the financial, creative and timing dimensions, a go-getter, with strong UX and design input.
Owned creative, content, deployment and localization on android.com, with meticulous capacity planning to hit targets across all workstreams without burnout. Exceptional communication and problem solving.
A year-long program at Arizona Public Service, complex and multi-part, with both project and client management. A team player and problem solver who built strong client relationships.
13+ years in program and creative operations. New York based. An independent page, not affiliated with Mercury.
Propose, then confirm
How I build AI into an ops workflow. A person approves before anything moves.
Built from zero
A creative operations function, defined and staffed by me.
6 teams
Six spreadsheet formats, thousands of cells, one resourced plan.
6 of 8
Property owners who gave up their own build process. That was the work.
01 · Point of view
Propose, then confirm
The same hour keeps going wrong. The resourcing meeting happens, real decisions get made out loud, and then somebody has to remember them and write them up afterwards. Usually late, usually thinner than what was actually said. The board drifts quietly out of date until the next meeting rediscovers it.
I built Vizor for myself, to manage my own team's capacity. The point of it is one sentence: the meeting should update the board, instead of somebody updating the board about the meeting.
Building it forced four decisions I would not have thought about from the outside, and those decisions are the actual point of view I would bring to this role.
The gate is the whole idea
Five steps, and the fourth one is a person. Nobody's week gets rearranged before somebody has looked at the change and said yes.
I care about this more than about which model anyone picks. A policy is a promise. Building the promise into the thing means nobody has to remember it on a bad Friday.
- 1Somebody says somethingin a meeting, or in a brief
- 2A card is draftedquoting the sentence it came from
- 3A person reads itand can disagree
- 4Approvethe gate
- 5The board updatesand not one step earlier
Every card quotes the sentence that caused it
A suggestion with no source is one you either swallow or ignore, and both are bad. Carrying the exact sentence back means a proposal is something you can argue with rather than something you have to trust.
The audio never leaves the room
The speech becomes text on the laptop, and only the text goes anywhere. You are asking a team to let software sit in on the hour where workload and staffing get discussed. That is the most sensitive hour in their week.
It runs on the team's own AI account
Encrypted, revocable, and they see their own bill. Without a key the AI features simply switch off and the capacity planning still works. Nobody is locked in by the clever part.
None of this is about AI being dangerous. It is about the resourcing meeting being the most politically loaded hour in a design team's week. Move somebody's work without them watching it happen and you do not get invited to a second meeting. Worth knowing: when Mercury shipped Command in June it landed on the same instinct, staging every action for a person to confirm. Different field, much higher stakes, same answer.
02 · The intake problem
Six teams, six formats, one plan
Every quarter, six teams at Dow Jones send in their plans and requests. Subscription strategy, product design, product management, performance marketing, CRM, event marketing. Each sends its own spreadsheet, in its own shape, half of it written in a hurry.
I started six months ago, so that has landed on my desk once. Once was enough. Asking everyone for a single template has never worked anywhere I have tried it, so I built Relay to take the files exactly as they arrive. The teams never open it.3
Click through the six steps. Watch the request on the right fill itself in, including the moment it admits it does not know something.
Six workbooks, no shared template
Nobody is doing anything wrong. Six teams have six ways of describing work, because they do six different jobs. The fix is not a nicer form. It is a thing that reads all six.
Six teams, thousands of cells. The volume is why this is a system and not an afternoon.
Requests do not arrive as tidy fields
They arrive as sentences in a cell. This is the one step where the AI genuinely earns its seat, because no formula gets you from that sentence to a resourced line item.
what was actually typed
"Q3 paywall test, need design + copy, maybe eng? mid Aug, check with Sarah"
Everything on the right is pulled out of a cell like that one. Nothing is re-typed by hand.
Six vocabularies, one set of terms
Three teams describe the same week three ways and the same page three ways. A plan built on six vocabularies is six plans, so everything downstream depends on settling on one.
- "mid Aug" · "8/15" · "before Labor Day"Aug 15
- "WSJ paywall" · "subs page" · "wsj.com/sub"WSJ subscription
- no owner named anywhere in the rowmissing
The third row is the interesting one. It does not guess. It flags the hole and moves to step four.
It chases its own missing pieces
This is the step I care most about. A tool that hands you a list of gaps has moved the work, not removed it. Relay writes the follow-up, sends it, holds the request open, and fills the hole when the answer comes back.
drafted, not written by me
"Quick one on the Q3 paywall test: who is the owner on your side? Everything else is in."
sentwaitinganswered
Nobody was chased by hand, and the hole from step three closes itself.
It proposes the ranking. I own it.
Every item carries a reason and a capacity cost, so the order is arguable rather than handed down. Anything low confidence or over capacity does not get quietly sorted. It lands in a tray that needs a person.
- NowQ3 paywall testdesign 3d · copy 1d
- NextBarron's event micrositedesign 5d · eng 8d
- Needs meMarketWatch refreshover capacity
Where the human stays. Relay does the reading and the arithmetic. Priority is a judgment call and stays one.
Nothing gets re-typed
Planning finishes in Relay. Only then does the finished plan move into Asana, where the PMs run execution and everyone sees their tasks. I own that Asana setup too, which is why the fields line up on both sides.
Relay
plan, capacity, sequence
Asana
tasks, owners, status
Plan in Relay. Execute in Asana. Never distort the plan to fit the tracker.
03 · What I built
Three tools, one method
Same three questions for each one, because the transferable thing is the method rather than the subject. The middle question is the one nobody can answer without having built the thing.
1 / 3
04 · The fleet
Nineteen agents, four squads, running on their own schedules
Nobody asked for this and nothing depends on it commercially. I built it on my own time, with my own work as the test case, because I wanted to know how far the idea actually goes. It is the most complete thing I can show you about how I take a problem apart.5
Command Herald, daily brief · Vault, finance · Counsel, legal · Ledger, weekly wrap · Pulse, heartbeat
Intelligence Scout, research · Atlas, strategy · Oracle, industry brain · Sentry, competitor watch
Growth Closer, sales · Compass, go to market · Beacon, marketing · Mercury, ads4 · Quill, content
Build Probe, QA · Forge, engineer · Frame, systems · Lumen, UI and UX · Prism, visuals
The unglamorous half
The same system's compliance view. Every open item sorted by risk and by product, across AI regulation, consumer protection, data privacy and marketing claims. Read the counts: mostly open, one cleared.
That is what a working tracker looks like in month one, and publishing it that way is the point. A screenshot with everything green would tell you nothing about whether it runs.
What matters operationally is that each agent runs on its own schedule, the hand-offs are traced, and the cost is tracked per agent. The work has a unit price and not just an output, so "is this automation worth keeping" has an answer instead of an opinion.
I am not an engineer.5 I write the requirements, direct the agents that build it, then review
and correct. What I bring is not code. It is knowing what to build, and having the
discipline to define it before anything gets built, because AI collapsed the cost of
making things and not the cost of thinking.
05 · The thesis
Where I think Design Ops needs to go at Mercury2
The posting asks for someone who defines the thesis rather than executing a list. So here is mine, built from what a careful reader can learn from your published work. Some of what I propose below will be wrong, because I have not seen the inside. Being told which parts, and why, is the fastest way I know to find out what I am missing.
Ship the system you already built, to the people who already need it
Your design tokens are readable in your own shipped stylesheet: a base palette feeding named, purpose-based tokens, each with its own hover, pressed, disabled and inverted state. The inverted set is why the marketing site can run dark while the product runs light off one source. That is careful work, and it has designers and engineers dedicated to it.
The Design Ops question is not whether the system is good. It is whether a designer in week one can find it, know which token to reach for, and know who decides when the answer is not in there. Design Ops does not own the system. It owns the distance between the system and the person who needs it right now.
Instrument the review, not the designer
The failure mode of design metrics is measuring output, which makes designers defensive and tells leadership almost nothing. In a growing design org the thing worth measuring is rarely how fast a designer works. It is how long work waits.
Intake×1.0
Spec×1.2
Design×0.9
Review×3.4
Handoff×1.1
baseline
This is not a theory I read. It is what I built into Vizor: moving work into review frees the assignee's capacity and starts a clock against that reviewer's own typical turnaround. Each reviewer is measured against themselves, not one org-wide number, because a legal review and a copy review are not the same promise. The measure implicates the system, not the person. That is the kind a design team will let you keep.
AI adoption is a workflow problem. Procurement is the thing that unblocks it.
The posting pairs owning the AI point of view with partnering with Procurement, and those belong together. Procurement is a gate, and gates are worth being good at, but it is not the bottleneck. Designers do not change how they work because a tool got approved. They change because two or three genuinely painful parts of their week got rebuilt with them, and the new version is obviously better.
Buywhen a product does the real job. Most of the time it does, and arriving wanting to build is the fastest way to lose Procurement's trust.
Adaptwhen a product does most of the job and the rest is process, not software. Change the process.
Buildonly when the near miss would force the team to reshape how it works around a tool's limits, and the build is days rather than quarters.
What makes this useful to a Procurement partner is not the framework. It is that it produces a written sentence explaining why the alternative was rejected, which is the thing they need and rarely get.
And what I would deliberately not do
A strategy without subtractions is a wish list
Not doing
A named framework
No capitalized methodology, no acronym, no rollout deck. In a design team this size, a named framework is a tax the team pays so Design Ops can be legible to people outside it. The team should be able to describe how work moves in plain sentences.
Not doing
Measuring designer throughput
No velocity per designer, no tickets closed, no output counts. Not privately, not "just for capacity." The moment a designer believes a number about them is being watched, the number becomes the work. If leadership needs an answer to "are we under-resourced," the honest one is wait time by stage, coverage against what we committed to, and a count of what we declined.
Not yet
Rebuilding intake in the first quarter
Every ops person's instinct on arrival is to fix the front door, because the front door is where the pain is visible. It is also the thing most likely to be holding something up in ways a new person cannot see. Read the queue for six weeks first, then touch it.
Not mine
The design system itself
There are already people whose job that is. Design Ops taking ownership of the system would be a land grab dressed as helpfulness. My job is adoption, documentation, the decision path when the system has no answer, and making sure the people who own it get the credit.
Learned the hard way
Ceremonies have to earn their slot
At Everyrealm I hired the team and introduced the production processes and ceremonies they worked inside, then cut the ones that stopped earning their slot. Adding process is the easy half and it is the half that looks like the job. Removing it is where the judgment is, and it is why rebuilding intake sits on the not-yet list.
06 · The posting, audited
Line by line
Sixteen lines from the job description. Nine I can prove outright, seven are partial. Open any line to see what I can point at, and every partial says exactly where the gap is. The partials are the reason to read the rest.
What the role owns
Set and run Design Ops' strategy, including what we are deliberately not doingPartial
Section five of this page is that thesis, with four named subtractions. Everyrealm supports it: I decided what a creative operations function would and would not cover, then staffed it. Marked partial because writing a thesis is an artifact, not a track record of having run one through execution inside a design org.1
Own Design's AI-first point of view, including building the skills, agents and workflows designers use day to dayPartial
Relay, Vizor and PACTO, built solo directing AI, plus a nineteen-agent fleet in four squads with traced hand-offs and cost tracked per agent. A stated buy, adapt or build framework I have actually used.3
Partner with Procurement to secure access to new toolsPartial
Vendor and partner management at Everyrealm and through Infosys on Arizona Public Service, plus agency-side work at Moving Brands and Prophet, so I have been on the selling end of the contract. My stated strength is profitability through resource utilization and capacity management.7
Architect systems that scale year over year without adding overheadMet
At Dow Jones, one design system replacing eight property-specific build processes, six of eight migrated, build time cut about 80%. At Vertic, a comprehensive design system for Arizona Public Service after a three-month research phase and an eight-month design phase, with Infosys and 28+ engineers.
Set and execute the design culture strategy: all-hands, onsites, team ritualsPartial
At Everyrealm I introduced production ceremonies into a team I had just hired, and cut the ones that stopped earning their slot. At Google I kept three web properties on one launch calendar across time zones.9
Build and run Design's external presence strategy, content to events to award submissions, with CommsPartial
Section eight is a researched read of Mercury's current external design presence and a concrete plan. Direct experience: android.com including the Android 12 launch, and the event and conference properties at Dow Jones, including JournalHouse and the WSJ Leadership Institute.6
Define and run onboarding and enablement, and set the quality bar for how new designers rampMet
At Everyrealm I sourced, hired and onboarded 3D artists, UI and UX designers, motion and visual designers into a function that did not exist before I arrived, and wrote the methodology they ramped into. The hiring bar and the onboarding bar were the same decision.
Be the cross-functional connective tissue across Product Ops, Research Ops, Eng Ops and design leadershipPartial
I am currently the bridge: brand steward for all product touchpoints and product-connected materials, fielding quarterly plans and requests from six functions, resourcing my team across design, content design and copy. At Everyrealm I worked across finance, architecture, product, marketing and web3 engineering with no playbook.10
Be a strategic thought partner to the VP of Design and design managersPartial
A year-long, deliberately ambiguous exploration for Google Search Ads, run in partnership with Google's Search UX Design Director and his team. At Arizona Public Service, bringing skeptical utility executives into the process rather than around it.11
Define what gets measured, and use it to make the case for where to invest nextMet
At GE Healthcare, the executive-level visualizations that kept a global campaign's leadership aligned while volume flowed underneath. At Dow Jones I own the Asana setup for reporting, capacity and resource management. Vizor is a committed position on what to measure: the wait, not the worker.
What the ideal candidate has
A track record of independently setting strategy for a horizontal function, defining the thesis rather than executing a listMet
Everyrealm: a horizontal creative operations function, defined and staffed from zero, spanning finance, architecture, product, marketing and engineering. I wrote the thesis and nobody handed me a list.
Break down silos and influence without authority across teams, product areas and platformsMet
Two years in Google's Android pod from Huge, leading a team of four to six designers on Huge's side and holding a peer relationship with Google's Search UX Design Director, with no authority of any kind. Four references on record, from a director, a Design Ops practitioner, an IC designer and an executive director.
Comfort making the call and driving decisions to a conclusion without full consensusMet
At Prophet, an MVP redesign shipped for a pharmaceutical client during the Endo–Mallinckrodt merger, where waiting for consensus was not on the menu.
A builder's mindset: owning strategy and execution alone, from concept to deliveryMet
Everyrealm from zero. Relay built rather than bought. Vizor and PACTO shipped solo. I do not have a maintenance résumé.
A high bar for quality that others point to, and accountability beyond your own workMet
Hulya G., Director of Web Marketing Strategy at Google, on the record about meticulous capacity planning across all workstreams without burnout. Gary Goldsmith, Design Operations at Meta, on the record about the quality of my program work.
Anticipating and mitigating risk across people, process and tools at department levelMet
Vizor's review clock exists because I watched projects slip on stakeholder feedback and wanted the slip visible before it became a miss. At Google, an immovable public launch date on Android 12.
07 · Adoption
Eight property builds, one system, and the part that was not engineering
This number is usually reported as an outcome: eight website properties unified into one, six of eight migrated, build time cut about 80%. Reported that way it sounds like an engineering result.
It was not. Consolidating eight property-specific builds meant getting agreement from the people who already owned those builds, and that half of it was not engineering.8
Two of the eight are not migrated yet. Six of eight is the honest state of it, and it is the number I would rather report than a rounded one.
I put this here because it is the fair challenge to everything in section three. Relay has one user and Vizor is a private beta, so if you want evidence that I can get other people to change how they work rather than build something they might, this is the piece of the record that carries it.
- Property 1
- Property 2
- Property 3
- Property 4
- Property 5
- Property 6
- Property 7
- Property 8
One system
6 of 8 migrated
~80% less build time
Eight event properties across WSJ.com, Barron's.com and MarketWatch.com. Unnumbered rather than named, because which ones migrated first is not mine to publish.
08 · External presence
What I found when I went looking for Mercury Design6
One of the responsibilities in the posting is owning how Mercury Design shows up externally, from content to events to award submissions. So I looked. What follows is what a search found. It is not a judgment about the team.
Found
- Found: Meridian, an editorial property with original commissioned photography and credited photographers. The most visible design investment.
- Found: The Art of Banking, a physical culture book, shown on the careers page.
- Found: Mercury Spheres, a company conference, inaugural edition May 20, 2025.
- Found: A mature internal design system, plainly visible in the shipped stylesheet.
- Found: A stated value: appreciate quality, pursue and recognize excellence to build something that lasts.
- Found: Deep, active design hiring across product, brand, systems, research and ops.
Did not find
- Did not find: Design-authored writing. The blog runs Library, Meridian and Inside Mercury. Product launch posts are written by PMs. No designer bylines surfaced.
- Did not find: Design awards. Fast Company Most Innovative Companies, Forbes Fintech 50, Fortune Best Medium Workplaces. All business awards. No Awwwards, Webby, D&AD or Communication Arts recognition surfaced.
- Did not find: A community footprint. No Dribbble, Figma Community or Medium presence surfaced for Mercury as a company.
"Did not find" is a statement about a search, not about the team. See footnote six.
The observation, and then the plan
A design team this size, with a token architecture that careful, a physical culture book and a commissioned editorial property, has more publishable material than most design teams that publish constantly. From the outside I could not tell who owns the pipeline that turns that material into published work. Your posting asks someone to own it, which suggests it is a job worth doing, so here is how I would start.
- First
One artifact, not a content strategy
The token architecture is the obvious first piece, written with the designer who built it and bylined by them, not by Design Ops. My job is the deadline, the editor and the publish button.
- Then
Treat awards as a calendar problem
Awwwards, Webby, Fast Company Innovation by Design and Communication Arts all have fixed deadlines and long lead times. A calendar with named owners, plus a habit of capturing process while the work is still warm, is most of the job. The craft is not the constraint here. The submission never getting written is.
- Measure
Hiring is the actual return
External presence is in this posting because a design team that keeps growing needs candidates who already know what the work looks like before they apply. So measure it there: source of applicant, quality of pipeline, time to fill. Not impressions.
09 · Selected work
The title kept changing. The job did not.
Thirteen-plus years of the same work under names that kept changing, because the function did not have a settled one.
Huge @ Google · 2020 to 2022
android.com, through the Android 12 launch
Roadmap, creative, content, deployment and localization across android.com, tv.google and wearos.google.com, against a public date that could not move. Separately, a year-long research and design phase for the future of Google Search Ads, leading four to six designers on Huge's side with Google's Search UX Design Director and his team.
Everyrealm · 2022 to 2023
A function and a team, built from nothing
A Web3 metaverse start-up that initially received $50M+ from A16z and others, later pivoting to game publishing. I sourced and built the team of 3D artists, UI and UX designers, motion and visual designers, defined the project plans, and introduced the production processes and methodologies. The function did not exist before I arrived. This is the closest thing in my record to a team of one setting direction and then doing the work.
Vertic · 2017 to 2020
A design system that had to survive 28+ engineers
A new transactional .com and mobile app for Arizona Public Service, built on a comprehensive design system, after a three-month research and strategy phase and an eight-month design phase, delivered with Infosys and a team of 28+ engineers. I structured the review cadence so misalignment surfaced before build rather than during it.
GE Healthcare · freelance
Executive reporting, at volume
The asset-production phase of the global Better Health Study, one of the company's largest marketing campaigns: executive-level visualizations, a global website, hundreds of media assets, and the supporting data analysis. The reporting is the part that transfers. Keeping leadership aligned on a cadence while the volume flows underneath is the same job as making a design function legible.
The Row · an Everyrealm project
Holding outside work to a standard
A members-only metaverse of one-of-a-kind 3D architectural landmarks by renowned artists. I owned the vendor relationship through implementation and platform testing to a single coordinated launch, plus external motion, 3D rendering and video production. Vendor management is Procurement's other half, and it is where I have most of my experience.
- 2013 to 2015Beyond, Digital ProducerGoogle's "Google for Work" site, on a tight deadline, with a team across time zones.
- 2015 to 2016POSSIBLE, Digital Project ManagerPetfinder.com redesign, rebrand and launch campaign. Purina Veterinary. ConEdison's new .com.
- 2023 to presentFreelance, Sr. PM and Digital ProductionGE Healthcare. Moving Brands, a UX and IA audit for NYU Langone's design system. Prophet, an MVP redesign during the Endo–Mallinckrodt merger.
- 2026 to presentDow Jones, UX Program Manager, BrandBrand steward for all product touchpoints across WSJ.com, Barron's.com and MarketWatch.com. The eight-into-one design system. The Asana setup. About eight events and five launches over five months. Built Relay.8
10 · On record
Quoted, not paraphrased
All four in full, including the two that rotate at the top of this page. A Design Ops practitioner, a director, an IC designer and an executive director. Four altitudes, on purpose.
Diligent on the financial, creative and timing dimensions, a go-getter, with strong UX and design input.
Owned creative, content, deployment and localization on android.com, with meticulous capacity planning to hit targets across all workstreams without burnout. Exceptional communication and problem solving.
A year-long program at Arizona Public Service, complex and multi-part, with both project and client management. A team player and problem solver who built strong client relationships.
Exemplary program management on a complex, large-scale project. Diligent planning, adaptability, and expectation management both internally and externally. His role was critical to its success.
11 · Roadmap
A plan whose output is "alignment" is not a plan
Design Ops is a team of one, so none of this is delegated. Each step names the thing that exists at the end of it.
- Weeks 1 to 2
Read the queue, not the roadmap
Every path work takes into design: where it enters, who bypasses the front door, and why they are right to.
Output: a map of how work actually arrives, including the routes nobody documented.
- Weeks 2 to 4
Sit in the reviews
Where does work wait, and on whom. Not to assign blame, to find out whether the waits are one reviewer, one stage, or one week of every month.
Output: a first, unflattering picture of wait time by stage, shared with design managers before anyone else sees it.
- Weeks 4 to 6
Audit the calendar and the tool stack together
Both are inventories of what the org has already said yes to. Neither gets read as an inventory until someone does it on purpose.
Output: a kill-or-merge list for rituals, and an honest read on spend, overlap and real adoption before renewal season decides for us.
- Weeks 6 to 8
Publish the prioritization logic
Written, observable principles, so priority becomes a lookup rather than a negotiation and the loudest requester stops winning by default.
Output: one page, in the open, that anyone can check my decisions against.
- Weeks 8 to 10
One AI pilot, small and real
Two genuinely painful recurring parts of the week, rebuilt with the designers who own them, with a measured before and after. Not an org-wide rollout nobody asked for. This one runs through security review and procurement, so week ten is the target and not the promise.
Output: a working thing two designers use every week, and a number.
- Weeks 10 to 12
Make the function legible, and give a piece of it away
One leadership view, one short changelog, one adoption number per surface. Each owned by someone other than me.
Output: a reporting cadence that survives my calendar, plus the first written case for what Design Ops should invest in next.
Not in the first 90 days
Reorganize anything, replace a tool, or introduce a framework with a name
All three look like leadership. All three spend the trust you need to make the boring changes that work.
12 · The ask
The job has had a lot of titles. It has been the same job.
Producer, project manager, creative operations lead, program manager. This is the first posting I have read that asks for all of it at once, including the part I built on nights and weekends.
pr.wvelazquez@gmail.com · 917-554-5173 · New York · LinkedIn · PACTO
Disclaimers and footnotes
Every asterisk on this page resolves here
Mercury footnotes its own headline, which makes it a company that has built qualifying its own claims into the brand. So this page does the same with mine. These are not fine print. They are the part I would want read first.
- 1.The step up. I have built the operating system for a creative team once from nothing, at Everyrealm, where I also hired the team. I am the operating layer of a brand function now, at Dow Jones. I have not held a role titled Design Operations, I have not run Design Ops inside a 40-person design team, and I have not reported to a VP of Design. What I have is the less common half: I have built the function rather than inherited a working one, and for a team of one, that is the half that matters. Back to the audit
- 2.The thesis. Section five is my read, written from outside, from public material only. I have not seen your roadmap, your research, your internal documentation, or where the work actually hurts. So some of what I propose there will be wrong. That is the point of writing it down rather than waiting: being told which parts, and why, is the fastest onboarding I can think of.
- 3.The AI enablement claim, which is the one to read. Relay has exactly one user: me. The six teams send spreadsheets. They never open it. Vizor is in private beta, invite-only, in real weekly use by me and by a few agencies run by friends, and it is not deployed across Dow Jones. PACTO is on private staging. So: I have built the agents, tools and workflows the posting asks for, and I have not yet rolled an AI enablement program out to a 40-person design team. That is the program I want, not one I have already run. The nearest evidence that I can move other people is section seven, which is about eight property owners rather than a tool. Back to section seven
- 4.The agent named Mercury. One of the nineteen, in the Growth squad, is called Mercury. It was named well before this posting existed.
- 5.Not an engineer. I write the requirements, direct the AI agents that build it, then review and correct, with code review from staff engineer friends at Meta and Google. I do not hand-write production code, and nothing here should be read as claiming I do. I am also not starting a business. These are proof of how I think about operational problems, not a venture with a future.
- 6.The external presence read. Section eight reports what a careful search of public material found and did not find, in July 2026. Mercury may well have an internal design blog, an awards calendar already in flight, or a deliberate decision not to publish. If so, that is the first thing I would want to be told, and it would change the plan rather than embarrass it. "Did not find" is a statement about my search, not about the team. Back to section eight
- 7.Procurement. I have managed vendors, partners and agency contracts from both sides of the table. I have not owned a formal Procurement and Finance cycle for a design organization's tool budget through a renewal season. That is a real gap and it stays marked partial. Back to the audit
- 8.The Dow Jones work. The brand hub my team's visual system feeds into is in progress and has not shipped, so nothing here should be read as a launched artifact. Relay is internal with no public address. Every Dow Jones figure on this page comes from my résumé: eight properties with six migrated, about 80% less build time, about eight events and five launches over five months.
- 9.Design culture at this scale. The gap is not the title. It is that I have run production ceremonies for a team I hired at a start-up, and I have not designed and run an all-hands, an onsite, or a distributed-team ritual programme for forty people with a budget and a Comms partner. Different craft, adjacent muscle. Back to the audit
- 10.Research Ops specifically. I have worked across nearly every adjacent function: product, engineering, marketing, finance, legal, brand, content design. I have not partnered with a dedicated Research Ops practice, because none of the organizations I have worked in had one. Product Ops and Eng Ops I can speak to from experience. Research Ops I would be learning. Back to the audit
- 11.Thought partnership to a design executive. The Google Search Ads relationship was a genuine peer partnership with a design director. It was a peer partnership, not the ops counterpart to an executive who owns a function, a craft ladder and a headcount plan. That reporting relationship is new to me. Back to the audit
- 12.Mercury's facts. Every Mercury figure and quotation on this page comes from Mercury's own published material as of July 2026. This page deliberately states no yields, rates or APYs: Mercury footnotes and dates every one of them because they move, and a personal page has no business restating a number like that.
- 13.The styling. Built with Mercury's published design language, from values readable in your own shipped stylesheet. No Mercury logo, no Arcadia font files, no copied code. The type is Inter and Inter Tight, chosen as free substitutes because they can be set at the same unusually light display weight that makes your typography read the way it does. Every capacity chart and diagram here uses representative values and says so.
- 14.Not affiliated. This is an independent page by Warren Velázquez. Not affiliated with, endorsed by, or connected to Mercury Technologies, Inc.