Prompt Engineer (Retell AI/ ElevanLabs) - Part-time - Worldwide

About the company

We are building AI agents designed to help small and medium-sized businesses (SMBs) grow revenue through smart automation.

We're seeking a Prompt Engineer to help shape and improve the brains behind our agents by deg, optimizing, and maintaining prompts that reliably drive useful, domain-specific conversations.

Role Overview

As a Prompt Engineer, you’ll own the design of high-quality prompt logic for LLM-powered agents. You’ll work closely with CEO and industry strategic partners to ensure our agents perform accurately, safely, and with contextual awareness.

Responsibilities

• Design, test, and iterate on LLM prompts for the core agent workflows (customer booking, re-engagement, marketing).
• Develop system and message prompts for multi-turn conversations using Retell AI, ElevenLabs Conversation AI Agents and OpenAI APIs.
• Collaborate with engineers to integrate prompts into deployed agent pipelines.
• Implement prompt-based workflows with dynamic inputs from structured data (e.g., CRMs, forms, calendars).
• Optimize for accuracy, latency, coherence, and output stability of the AI conversations.
• Build safety guardrails into prompts to minimize hallucinations, and inappropriate outputs.

Qualifications

• Experience working with LLMs such as OpenAI GPT-4, Claude, Gemini, or open-source models.
• Strong experience crafting effective prompts for multi-turn interactions, and tool use.
• Experience integrating prompts into agent orchestration systems with Python or Node.js.
• Proven ability to evaluate and iterate on prompt performance using metrics and qualitative feedback.
• Strong familiarity with Retell AI and/or ElevenLabs or similar voice-based AI tools for real-time AI conversations.

Preferred Qualifications

• Hands-on experience with prompt engineering frameworks or libraries
• Understanding of SMB operations, tools (CRMs, schedulers, forms), and automation pain points.
• Ability to debug LLM outputs and improve reliability using structured prompting or dynamic inputs.