Currently · how I'm running my own search

How an operator runs her own job search. With receipts.

I'm an operations leader between roles. I built this dashboard to keep myself honest about what's actually working and what isn't. I only apply to roles that are great fits. Real numbers, real lessons, straight from my tracker.

Updated: 2026-06-09 Open to work Jacksonville, FL · Remote US
The dashboard

The live state of my pipeline.

What I'm applying to. How it's converting. Where the pipeline's getting stuck. The numbers come straight from my tracker. No company names here, just the shape.

The conversion flow

From application submitted to offer accepted. Numbers on the arrows show how many applications took each path.

Roles by category

The shape of what I'm targeting. AI-coded ops, BizOps, and Strategic Ops lead the mix.

Application status

Applied / Rejected / Evaluated. The rejection rate is the data point driving the pivots below.

Fit score distribution

Every role is scored against my profile (proptech anchor, Jacksonville/remote, no sales-adjacent, etc.) before tailoring time gets invested.

Time to first response

Days from apply to recruiter response. The 3-day median tells me most no's never reached a human.

My daily routine

Four steps that run every day.

Job searching is mostly a momentum problem. I built a daily routine so I have one whether I feel like it or not. Some of it runs in code, the rest I do by hand.

Step 1

Scan

Daily 6am remote agent. Three passes: 45+ startup ATSes via direct API, 21+ Jacksonville-area companies, 4 LinkedIn lanes. Liveness re-check on every fit before adding to the log.

Step 2

Tailor

Resume and cover letter rewritten for each role. The JD's language gets borrowed only where it actually grounds in my work. Featured project framed to match what the role values.

Step 3

Apply

Same-day on fresh postings whenever I can. Through the company's own portal, not a job-board aggregator. Gaps acknowledged head-on in the cover letter, never hidden.

Step 4

Follow up

+7 days for first check-in, +14 days for final. Through the application channel, never cold LinkedIn DMs. Track every response in the master tracker.

What I've learned so far

The honest read.

Updating this as I go. Some of these wins surprised me, and some of the misses still sting. Both are worth keeping in writing.

▲ Working
  • Warm intros land materials in front of a human. Routing through a former boss or family connection bypasses the ATS filter and the recruiter triage stack. Highest-leverage move in the search.
  • Same-day applications on fresh postings. Stack timing matters. A role posted Friday morning has a dozen applicants by lunch and a hundred by Monday.
  • Cover letters that quote the JD verbatim. Mapping my exact work to the JD's exact asks signals tight read of the role, not a copy-paste application.
  • Acknowledging gaps directly. Tenure, industry, or scope mismatches addressed upfront. Recruiters respect the honest framing more than they reward a polished application that ducks the question.
  • Treating the portfolio site as the second screen. Recruiters who click through land on a Featured Project with a live prototype, an analytical brief, and a Loom walkthrough. The CV is the door; the portfolio is the room.
▼ Hasn't
  • Cold applications generally. The signal-to-noise ratio is low. Most rejections come in 1-3 days, which means human review never happened. Warm intros are the obvious antidote.
  • The healthcare domain bet. A coordinated push into healthcare ops in mid-May converted at zero. Proptech is the stronger domain anchor by a wide margin.
  • Pulled roles between scan and apply. Lost a couple of strong fits to mid-day postings being yanked. Liveness re-check is now mandatory before any tailoring time gets spent.
  • LinkedIn search behind the login wall. Free-tier results don't render to anyone who isn't logged in, which means my morning scan can't see them. I check those by hand.
  • Cold LinkedIn DMs as “warm intros.” Not the same thing. Real warm intros work. Cold DMs feel invasive on both sides.
How it's wired

What runs underneath.

AI is part of the toolkit, not the headline. The real work is the cadence and the metrics. The hard part is changing course when something isn't landing.

Claude ChatGPT Node.js Custom scan agent on cron Greenhouse / Lever / Ashby APIs Google Drive sync via rclone Markdown source of truth CSV + Sheet mirror Playwright for PDFs Schema-first prompts