Full Time
$10 - $25 USD/hour
TBD
Apr 18, 2026
If you work with Claude Code, Cursor, Aider, or similar AI coding tools every day — and you catch what they get wrong instead of merging bugs — we want to talk.
## What we do
Joburn runs 10+ production Next.js apps for agencies, creators, and coaches. Supabase backend. Vercel hosting. Direct API integrations with Claude, Gemini, HeyGen, and other LLM providers.
We use AI coding tools non-stop. We ship faster than a small team has a right to. But the founder is the bottleneck on review and integration. That's where you come in.
## What you'll do
- Take specs from Claude Code, finish them, harden them, ship them
- Build LLM-powered features end to end: prompting, tool use, streaming, evals
- Own Supabase schemas and RLS for multi-tenant apps
- Work across 10+ apps in our ecosystem
- Debug what AI generated but didn't understand
## What you need
- 5+ years production Next.js, React, TypeScript
- Claude Code, Cursor, Aider, or equivalent as your daily driver
- LLM APIs in production. OpenAI, Anthropic, Gemini. You've hit rate limits. You've debugged tool calls. You've built evals.
- Supabase, Postgres, RLS policies from scratch
- Security discipline. You audit AI-generated code before merging.
- Strong async English
Bonus if you know:
Double bonus if you know: Jobs To Be Done (JTBD), The Mom Test, Switch framework, or similar product-thinking methodologies. We think in outcomes, not features — if you do too, we'll move very fast together.
## The deal
$2,500 - $4,000 USD/month, 30-40 hrs/week. Biweekly pay via EasyPay.
Paid trial project first (2-3 days, $
First engineering hire at a company scaling multiple AI products. Growth path into Lead Engineer as we build the team.
## This role is for you if
- You think from first principles. You understand the problem before you write the code. You know that a well-defined problem is 90% of the solution — the rest is wiring.
- You're proactive at an elite level. You don't wait to be told what's broken.
- You've shipped production code that real users depend on — not demos, not templates, not agency throwaways
- You catch what AI coding tools get wrong instead of merging their bugs
- You treat security as a constant, not a feature request
- You want to stick around for 2+ years and grow as we grow
- You read job posts carefully (and you'll follow the subject line rule below)
## This is NOT for you if
- You just execute specs without asking why — "why" is the most important question you'll ask here
- You've never written an RLS policy from scratch, or you think RLS is the same as middleware auth
- You've never used Claude Code, Cursor, or a similar AI coding tool in real work
- You're looking for 3 months of work then moving on
- You're an agency, team, or staffing firm bidding on behalf of someone else
- You write proposals with ChatGPT and don't read job descriptions carefully
## How to apply
Change your
Example: JOBURN-ENG | Python
Then answer these 3 questions inside your application:
1. Paste an RLS policy you wrote. What table, what rule, why it exists.
2. What kind of work makes you lose track of time? Doesn't have to be code — could be anything.
3. When you hit a wall on a hard problem, what do you usually do?
Wrong subject line format = auto-reject. Generic ChatGPT-style answers = auto-reject. We read every word.
## About Joburn
Joburn Pte. Ltd., Singapore-registered. We build AI-powered growth infrastructure for SaaS companies, agencies, creators, and coaching businesses. We've helped clients generate multi-seven-figures in revenue.
Small distributed team. We move fast, ship quality, and get shit done. No corporate politics, no committee meetings, no busywork.