30-Second Read

Key Decisions

Drew the safety line before the feature line: URL-only on restricted platforms, no scraping, no email guessing, no auto-send — the user always reviews and sends. Trust over convenience.

AI · Data Logic

Chose deterministic parsing + evidence-grounded contact discovery over an LLM black box for the MVP, so every output traces to a public source — then scoped a clear upgrade path to an LLM drafting layer and learned ranking once the rule-based loop is validated.

Result

Shipped an end-to-end 8-step outreach workflow from PRD to a packaged Chrome extension, turning a multi-tab manual chore into one structured, evidence-backed, copyable flow.

What I Learned

Designed the evaluation around contact-evidence precision, parser coverage, and draft usability — and learned that for trust-sensitive AI tools, the guardrails are the product, not a constraint on it.

LaunchedChrome extension · Workflow automation · Job-search outreach

Recruiter Finder Extension

A Chrome Side Panel outreach workflow that turns public job postings into recruiter evidence, resume-aligned highlights, and copyable drafts.

Role

AI Product Manager

Timeline

May 2026 – Jun 2026

Core Problem

When job seekers find a suitable role, they switch between the job page, company site, LinkedIn, Google, their resume, and outreach drafts to decide whom to contact, whether a public email exists, and how to write a concise, role-specific note. It is slow, error-prone, and risky when tools scrape restricted platforms, guess emails, or auto-send. The core problem: turn a public posting into verifiable, copyable, user-reviewed outreach without scraping restricted platforms, fabricating contacts, or auto-sending.

Target Users

Job seekers who want structured, safe outreach instead of cold mass-applying — especially candidates for product, project management, technical program management, and engineering roles.

Key Results

—Planned and shipped an 8-step job-to-outreach workflow: page detection → official job parsing → contact-evidence discovery → resume profile → job–resume matching → recruiter-search fallback → ≤300-char LinkedIn note → cold-email draft.
—Turned multi-tab manual job outreach into one structured, evidence-backed, copyable 8-step workflow — scoped as 19 requirements and 73 issue-level tasks across parsing, contact evidence, confidence states, safety guardrails, resume matching, outreach generation, QA, and rollback.
—Built a local-first MVP: Chrome Manifest V3 Side Panel + vanilla HTML/CSS/JS frontend + Node/Express local backend exposing health, official job parsing, contact discovery, and resume parsing.
—Established safety boundaries: URL-only mode for LinkedIn / Indeed / Handshake; no DOM scraping, login-wall bypass, email guessing, auto-apply, auto-send, or auto-LinkedIn actions.
—Delivered verifiable engineering: 41 extension/backend/test source files, 5 core test files, and a packaging script that produces dist/recruiter-finder-extension.zip.

⌥GitHub↗