Any
Up to PHP 30,000/month
40
Sep 25, 2026
Build the harnesses that let AI agents actually do work: tools, orchestration, guardrails.
ABOUT US
We run content, online coaching and commerce operations powered by AI agents that use tools, browse, call APIs and coordinate. You will build and harden the harness around them: the tool layer, orchestration, context management, and the guardrails that keep them on task.
We are hiring for multiple positions. Full-time and part-time both welcome.
WHAT YOU'LL DO
- Build and extend agent frameworks: tool definitions, tool-use loops, multi-agent orchestration
- Engineer context and memory so agents stay coherent over long tasks
- Add evals and guardrails; catch and contain an agent going off the rails
- Integrate agent systems with our real tools and data
MUST HAVE
- Deep Python
- Hands-on LLM tool-calling / function-calling
- Experience building or heavily extending an agent system
- Understands prompt and context engineering as engineering, not vibes
NICE TO HAVE
- Claude Agent SDK / Claude Code
- MCP (Model Context Protocol)
- Eval frameworks
- Experience with unreliable-tool retry and recovery
PAY: Up to PHP 30,000/month (full-time), based on skill. Part-time pro-rated.
HOW WE HIRE
Screen on the "To apply" ask below, then a short PAID test task, then a small contract, then ongoing.
TO APPLY
Show an agent or tool-use system you built. Tell us one time it did the wrong thing and exactly how you stopped it from happening again. Applications with only a CV are skipped.