Job Finder — the hunt as a product

Not a spreadsheet of applications — a command center for search, prep, and honest momentum.

The problem

Job search is emotional and operational at once. Tabs multiply. Prep gets lost. Outreach without a protocol becomes noise — or silence.

I built Job Finder because I needed the hunt to feel like something I could steer: a place where tools, agents, and my own progress share one map.

The approach

Job Finder is the hub. Extensions, Nexus governance, LinkedIn draft tooling, and prep flows plug in — they do not replace it.

Design commitments:

  • Agents may draft and explore; they do not send without recorded human approval.
  • Public narrative defaults to documenting decisions; private stays private unless marked otherwise.
  • A “player” path and local profile make progress tangible without turning the search into a game that lies.

The craft

  • Nexus alignments that bind the whole hunt stack.
  • Chrome relay patterns that drive my logged-in browser — cookies preserved — with assist on, publish still gated.
  • Interview and workshop surfaces that treat prep as buildable craft, not last-minute panic.
  • A shared player profile so identity, wallet, and progression can travel across related tools.

Experience context

Before the solo agent-OS years: Machine Learning Developer at CEVA (2018–2022), building and optimizing ML / algorithmic components in a DSP environment — prototype through implementation. That industrial rigor still shapes how Job Finder treats gates, traces, and “nothing ships without a human.”

Want more?

I will demo the hub and the gate protocol for recruiters or hiring managers who want to see how the pieces fit. Selected code by request.

Skills in play

  • Full-stack product systems
  • Agent-assisted workflows
  • Interview / prep tooling
  • Local profile and progression design
  • Browser automation with human gates

Talk to me All stories