Solo-shipped product · live
HiddenJobs.pl
AI-matched 100% remote IT roles for Polish specialists — personalized over offers from foreign companies validated for collaboration with Poland. Live, with real users, shipped solo.
The story · Feb → May 2026
Feb 2026
Started with validation, not code
Before writing a single line, I ran demand validation. Talked to specialists, asked what was broken about the existing job boards, watched what people actually needed vs what they said they wanted.
Core insight: senior specialists don't need more job offers — they need the rightones. The thesis I built around: source offers from foreign companies that actually hire Poland-remote, then AI-match each one to the specialist's CV + natural-language preferences. Curation + personalization, not aggregation.
Mar 2026
Built the MVP after validation locked
Started shipping. The core mechanic: CV analysis (OpenAI) + natural-language preferences → ranked offers per specialist. Backend infrastructure for sourcing offers from foreign companies, AI-assisted extraction so listings stay structured and clean, Poland-collaboration validation per offer, the public job board, the per-specialist matching surface. Stack: Next.js 16, NestJS, PostgreSQL/Supabase, OpenAI.
This is the part that historically would have stalled me. With AI agents in the loop, it didn't.
Apr 2026
Public launch · waitlist filling fast
Public job board released. LinkedIn launch posts went out the same week. Waitlist crossed 120+ in the first weeks; 40 specialists converted to active accounts.
May 2026
Content engine + first AI search citations
Started a long-form content engine in Polish — 9 thematic tracks planned, posts published bi-weekly. Within a month: 5,000+ article views, 100+ outbound clicks, and the first ChatGPT referrals showing up in analytics on day 7.
For a brand-new domain, getting cited by AI search within the first week is the validation signal that the content engine is working.
What I built — solo, end-to-end
Public surface
Next.js 16 App Router job board with ISR-driven listing pages, structured listings, and a content engine.
Backend pipeline
NestJS API + CLI tooling for sourcing, AI-assisted extraction, and a clean data model. PostgreSQL via Supabase, deployed on Render.
Specialist surface
Sign-up, CV upload + AI analysis, natural-language preferences, the per-specialist matching surface where each offer is ranked against profile, click tracking, an admin panel I run the whole product from.
Operations
Analytics (PostHog), search console integration, email automation. One person owns all of it.
What “AI-augmented engineering” meant in practice
Three real shifts that wouldn't have been possible two years ago.
1. The 60% wall stopped existing.
Before, my own products would consistently hit a wall around 60% complete — the part where you have to build the boring infrastructure (auth, DB schemas, deployment, content management). Solo, after an 8-hour day job, you don't have the willpower for that. With agents in the loop, that part stayed in motion.
2. The bottleneck moved to product judgment.
The hard question stopped being “can I execute this” and became “is this actually worth building.” That's the right place for the bottleneck to be — and it's the question I'd been trying to ask for years.
3. Operating a product as one person became a real option.
Running HiddenJobs day-to-day means doing sourcing, content, SEO, analytics, support, and engineering in parallel. None of it is on autopilot. But all of it is now in the realm of what one engineer can hold — if the engineer treats agents as co-engineers, not as autocomplete.
This is the working mode I'd like to bring to a founding team.
If you're an early-stage startup where AI-augmented engineering is the default — let's talk.
About me