Product Engineer/Tech Lead - Agentic Recruitment OS

Product Engineer/Tech Lead (Fullstack) - Agentic Recruiting OS

Location: Remote (preferably any timezone with ? 4 h overlap to CET)
Engagement: Full-time, 40 h/week
Compensation: USD 900 - 2'300 base/month + Performance Bonus (USD 100 – 250 per successful placement made through the system you ship) + path to equity/co-founder stake for the right person
Start: ASAP

Intro
I'm a headhunter. I built a 100 000-line AI-native CRM/ATS that is already running my own agency. Now I'm turning it into the first true Agentic Operating System for recruiting, a SaaS where customers don't subscribe to software, they subscribe to a digital workforce.
The MVP isn't a prototype. It's almost production-ready. I need a technical partner to take the wheel: architecture, multi-tenancy, agent orchestration, customer onboarding, the works. You will be the technical lead from day one, no PMs above you, no Jira theatre, no committees.
If you are a "9-to-5, Slack-status-yellow" developer, please don't apply. If you've shipped a SaaS, live in Cursor/Codex/Claude Code and routinely have 5+ agents running in parallel, keep reading.

What's already built
Not a green-field gig. The foundation is real, battle-tested, and producing revenue:
• React 18 + TypeScript frontend — Vite, Tailwind, shadcn/ui, 5 languages, full PDF pipeline
• Supabase backend — 147 migrations, 36 tables, 172 RLS policies, role-based access control
• pgvector + HNSW cosine indices for semantic candidate ? job matching
• 48 Edge Functions (~18k LoC) — each effectively a ready-to-wire agent tool: CV parsing, predictive market radar, deep-reasoning seniority engine (L1–L6), employer-conflict detection, exposé generation, public job feed, full ATS flow + ~35 more
• Live in production today — full Candidate / Client / Job / Pipeline / Recruiting / Tasks / Applications management, AI Matching Engine, Predictive Market Radar, Career-Page backend, full CV pipeline
Quality bar: TypeScript end-to-end, generated DB types, RLS-enforced security. Not perfect — but a working foundation.
The Mission, in three phases
1. SaaS-ify the Beast — multi-tenant SaaS with org/workspace isolation, Stripe billing, per-tenant secrets, hardened RLS
2. Agent Orchestration — wrap the 48 Edge Functions in an agent runtime (your call: LangGraph, OpenAI Agents SDK, Mastra, Inngest, custom). Long-running, queue-backed agent loops. MCP host so customers plug in their Slack / Calendar / Wiki / ATS (BYOT system) and our agents act on their behalf
3. Scale & Customer Wins — migration pipelines from Bullhorn, Vincere, Pipedrive, Recruitee, Excel. Multi-provider LLM routing (OpenAI + Anthropic + Gemini) with evals, fallbacks, and cost controls. WhatsApp + email outreach automation
You own the product architecture and the technical roadmap. I bring product vision and customer pain — you bring the tech direction. Once we hire the next engineer, you lead them
Who you are
• Cracked builder. Shipped real products, ideally as a founder or first engineer. You don't need a ticket to know what to build next
• AI-native. Cursor is muscle memory. RAG, function calling, vector search, LLM evals — strong opinions on which model fits which task
• Strategic thinker. Ship fast where it doesn't matter; obsess over fundamentals where it does (security, multi-tenancy, data integrity)
• Proactive communicator. When my idea is wrong, you tell me — with reasoning, not opinions
• Bonus points: prior SaaS, Supabase + Edge Functions in production, built an agentic system before, speak German, or have HR-tech context
Why join
• Full ownership. Technical lead from day one. No micromanagement.
• Strong long-term upside, discussed individually with the right candidate after a few weeks of working together
• You win when I win. Performance bonus tied to every placement my agency closes through your system, every month, before SaaS revenue starts
• Real distribution. I'm the first paying customer. Warm pipeline of befriended agencies ready for closed beta
• Coding-agent-native War Room. Multiple parallel agent instances are how we ship — no need to convince anyone AI-assisted dev is legit
• No-bullshit timeline. Closed beta this summer, public launch in autumn, fundraising thereafter (2027)
Stack - open to evolution
The current stack got us to production. It is not sacred. We are explicitly open to a different stack and architecture if you can argue why it makes us faster, more scalable, or more cost-efficient.
Today: React + TS + Vite + Tailwind + shadcn/ui · Supabase (Postgres, Auth, Storage, Deno Edge Functions) · pgvector + HNSW · RLS-based access control · single LLM provider with embedding pre-selection + caching · commercial search/scraping APIs + Google Maps · custom Postgres job-state pattern.
You'll build, evolve, or replace: multi-tenancy + Stripe billing · agent orchestration runtime with observability + human-in-the-loop · MCP host · multi-provider LLM routing · WhatsApp + email outreach · possibly Next.js for the public side · migration pipelines.
Non-negotiables: solid security, clean multi-tenancy, shipping speed. Everything else is up for debate.