What you'll do
- Build agents with the Laravel AI SDK and Claude or OpenAI models
- Design tools, prompts and evaluation sets, and measure what works
- Add guardrails: scope limits, human approval, logging and cost caps
What you'll need
- Production experience with LLM tool calling and structured outputs
- Solid PHP or Python, and comfort reading a codebase you didn't write
- A habit of measuring before claiming
How we work
Fully remote, on contract, with a fixed overlap window with clients.