JobMirror

Three weeks ago I quit overthinking and started building.

I've spent 10 years in HR, watching candidates make career decisions with nothing but gut feeling and a messy spreadsheet. I knew there had to be a better way. So I built JobMirror — an AI-powered platform that helps job seekers analyze job fit, optimize their resume, compare offers, and get a comprehensive career report.

Here's what happened.

What I built

Five core features, all in one flow:

• JD Fit Analysis — paste a job description, upload your resume, get a match score + gap analysis

• Resume Review — role-specific suggestions, not generic advice

• Personality Assessment — Big Five, mapped to career implications

• Offer Comparison — weighted scoring across salary, growth, culture, benefits

• Career Report — everything combined into one exportable document

Tech stack: Next.js 14, Cloudflare Pages (edge runtime), DeepSeek for AI, PayPal for payments.

The honest numbers

• Launch date: March 2026

• Users so far: early days, still growing

• Pricing: Free tier + $9.9/mo Pro

What worked

Starting with the problem, not the solution. I didn't build features I thought were cool. I built the exact workflow a job seeker goes through: find job → check fit → fix resume → get offer → compare offers → decide.

Shipping fast. The first version was ugly. I deployed it anyway. Real users found bugs I never would have caught in testing.

What was harder than expected

Edge runtime constraints. Deploying on Cloudflare Pages means everything runs on the edge — no Node.js APIs, no Prisma, no SQLite. I had to rewrite auth twice (next-auth v4 → v5 beta) before it worked.

Getting first users. Building is the easy part. Distribution is the real challenge. I'm still figuring this out.

What's next

• More blog content targeting long-tail SEO keywords

• Product Hunt launch

• Reddit and community outreach

If you're job hunting or know someone who is, I'd love your feedback: jobmirror.app

And if you've been through the distribution grind — what actually worked for you?