Chapter 14: Not a Rescue
I replied to the email before I could talk myself into making the answer sound smaller.
I have concerns, and I’ll raise them in the review. I can lead the discussion, but I won’t present unresolved questions as settled.
The response came a few minutes later.
That’s exactly why we asked you.
On Monday morning, I arrived at Meridian with the onboarding schedule folded in my bag, though every detail was already on my phone. The building was quieter than I expected. Nobody stared at me as if they recognized the girl from the graduation video. Nobody mentioned my sister, the torn letter, or David Mercer’s announcement.
At reception, I received a badge, a laptop, and a stack of forms.
My manager, Elena Ruiz, met me after security training. She walked me through the systems team’s priorities, then showed me where I’d be working. There was no grand office or instant authority attached to the title. Just a desk, a half-empty whiteboard, and a list of systems I would need to learn.
“You’ll have people to ask,” she said. “We expect you to listen before you make changes. And if something doesn’t make sense, say so.”
I liked her immediately for not pretending that confidence meant knowing everything.
That afternoon, I led the reliability review. The prototype was not on anyone’s screen. We discussed documentation, failure reporting, and what would have to be independently verified before anyone proposed an evaluation. I asked the team to distinguish three things that had been getting blurred together: what we had observed, what we suspected, and what we had not tested.
A senior engineer leaned back in his chair.
“That’s a lot of caveats for a system that isn’t even in use,” he said.
“That’s why it’s a good time to write them down,” I replied. “Before anyone assumes we know more than we do.”
Elena didn’t rescue me from the silence that followed. She let the room sit with it.
Then Priya opened her notes and added a question to the review list.
By the end of the meeting, we had more questions than answers. I left feeling better about the work than I would have if everyone had nodded and called it done.
My job was real. It had been real before the ballroom doors opened, before the CEO arrived, before anyone filmed the scene. The salary and start date were in my signed agreement. My first assignment had come from a calendar invite, not a dramatic announcement.
When a colleague asked whether the graduation video had changed anything, I told him the truth.
“It got attention. It didn’t get me hired.”
He nodded, then asked if I knew where the extra badge lanyards were.
I was grateful for the ordinary question.
The project did not move as quickly as the conversations around it might have suggested. Meridian and the independent reviewer needed to settle the scope, the safeguards, and the reporting rules. The revised terms made clear that any evaluation would use synthetic data in a test environment. No clinic records. No clinical reliance. No implication that the system was ready for use.
I read the terms twice, asked for one sentence to be tightened, and signed only after the language reflected what the team could actually promise. The evaluation was then approved within that limited scope. Approval did not make the prototype reliable. It meant we had permission to investigate a specific set of questions under specific conditions.
That distinction mattered.
For the next few weeks, I divided my time between my systems role and the project’s scheduled review work. My regular assignments were separate. If I left the project team tomorrow, the systems team would still have work for me. If the project never passed another review, my job would not disappear with it.
I began to understand how much comfort there was in a boundary written plainly.
The first round of controlled evaluation produced mixed results. Most recovery states matched what the team expected. One did not. After repeated interruptions, the record could return in a way that made a missing value difficult to distinguish from a value that had never been present.
No one called that a success. No one called it proof the whole prototype was useless, either. We recorded the result, suspended that test path, and asked the independent reviewer to examine the conditions that produced it.
The clinic continued to wait.
That was the right answer, even when waiting felt like a slow kind of disappointment.
At home, my family did not become a different family because I had a badge from Meridian. My mother sent a message about a Sunday dinner. I agreed to come for an hour. My father asked whether I wanted him to pick me up; I said I’d drive myself.
Vanessa didn’t ask me for a referral. She didn’t ask me to rewrite her résumé or explain how to handle an interview. Instead, she texted one afternoon to say she had made a list of marketing openings and planned to submit applications through the companies’ websites. She asked if I’d be comfortable meeting for coffee the following weekend.
I read it twice, trying not to confuse a good first step with a finished change.
I answered that I could meet for coffee. I added that I couldn’t manage her job search for her.
She replied, I understand. I’ll handle it.
I didn’t send reassurance. I didn’t tell her I was proud of her. I let her words stand on their own, and I kept my Saturday open for myself, too.
A few weeks after my first day, the independent reviewer completed a preliminary assessment of the recovery issue. The report didn’t offer the clean ending everyone wanted. It narrowed the circumstances that needed more investigation and listed the evidence still missing. The team used it to revise the test plan, then ran another controlled round within the approved scope.
At the final review, we sat around the same conference table where I’d first asked everyone to separate observation from assumption. A chart filled the screen: most of the tested recovery states were documented; the repeated-interruption case remained inconsistent under one specific condition.
Priya summarized what we knew. The reviewer described what could be verified and what still could not. Elena asked whether the team had enough information to recommend the next step.
The possible next step was another limited round of testing, with the unresolved behavior clearly marked and no path to clinical reliance. It was not a launch. It was not a promise that the problem would be solved. It was simply a proposal to continue investigating, under controls stricter than the prototype itself.
Everyone turned toward me.
“You’ve reviewed the revised scope,” Elena said. “Is that limitation acceptable for this next evaluation, or should we stop here?”
I looked at the chart again. The clean-looking parts of the display were not the same thing as proof of recovery. I had learned that in a quiet room, over weeks of careful work. I had learned it at my kitchen table, too, in a family that wanted apologies to mean more than they could mean on their own.
May you like
Acceptable for what?
That was the question I needed to answer before I let anyone move forward.
