Prompt Engineer / Conversational Designer

Moonfire; Austin (TX); Hybrid; USD 90,000 - 130,000; Full time

Job summary

Moonfire in Austin, TX is seeking a prompt engineer to shape how our AI talks to people across voice and text agents. You will design conversations, write prompts, and test them against real recordings and transcripts, owning prompt quality across products.

You'll work with product managers and engineers, iterate quickly, and ensure tone, clarity, and trustworthiness in eldercare-focused interactions. A hybrid, in-office schedule in downtown Austin is expected.

Qualifications

• Based in Austin and able to be in our downtown office a couple of days a week.
• Roughly 1-2 years hands-on prompt engineering.
• Evidence you've designed something, not just assessed it.
• Enough technical ability to build tools and discuss integration with engineers.
• Strong writing and communication skills.
• Familiarity with LLM APIs and conversational frameworks (e.g., LangChain).
• Excellent writing and linguistic skills, with attention to tone, clarity, and cultural sensitivity.
• Basic programming skills (Python preferred) for testing and prototyping prompts.
• Strong cross functional partnership and communications skills.

Responsibilities

• Design multi-turn dialogue flows for voice and text agents with context awareness and memory handling.
• Build, version and maintain a real prompt library - system prompts, flows, reusable templates.
• Test against real recordings, transcripts and simulations, before and after launch; hunt hallucinations and token waste.
• Partner with product managers and engineers to land prompts inside technical constraints; support compliance and privacy obligations.

Skills

Prompt engineering

Conversations design

Testing prompts

Python

LLM APIs

Cross-functional collaboration

Clear communication

Language and tone

Prototyping prompts

Tools

LangChain

OpenAI API

Anthropic API

VAPI

Job description

Description

About the Role

You'll help design the way our AI actually talks to people.

Sensi runs a growing family of agents - voice, text and summarization - across multiple product lines serving home care agencies, seniors and their families. This role owns the prompts behind them. Not one agent, and not one narrow surface: you'll build new agents from the ground up for new use cases, and you'll own prompt quality across several products and several product managers at once. It is a genuinely broad, dynamic remit for an individual contributor, and you'll have an experienced prompt engineer alongside you rather than being the only person thinking about this.

Concretely, that means writing and rewriting the prompts, designing the conversations, testing them against real recordings and transcripts, and iterating on what you find. You'll work directly with product managers on what an agent needs to do, and directly with engineers on how it gets built.

The work is high-stakes in a way most prompt work isn't. Our users are older adults and the people who care for them. Tone, clarity and trustworthiness aren't polish here - they're the product.

This role is based in Austin, TX, with a hybrid work setup, combining in-office collaboration with remote flexibility.

Key Responsibilities

Design the conversations.

Build multi-turn dialogue flows for voice and text agents, with context awareness and memory handling. Define each agent's personality and tone, and keep it consistent as the family of agents grows.

Build the prompts, and build the tooling around them.

Write, version and maintain a real prompt library - system prompts, flows, reusable templates. Where you need a harness or a script to test something properly, you build it.

Test against reality, not vibes.

Work from actual recordings, transcripts and simulations, before and after launch. Hunt hallucinations, wrong tone, dead ends and token waste in live interactions, then close the gap.

Partner across the org.

Take requirements and priorities from product managers; work with engineers to land prompts inside real technical constraints; support our compliance and privacy obligations. Share what you learn - this is a new enough discipline that your findings change how other teams work.

Requirements

What we're looking for

• Based in Austin - able to be in our downtown office a couple of days a week. We work closely with colleagues in Israel, so real overlap with both Austin and Israel hours matters.
• Roughly 1-2 years of hands-on prompt engineering - genuinely writing, building and iterating on prompts yourself. This is a young field and we're not looking for a decade of it.
• Evidence you've designed something, not just assessed it. The distinction matters to us: rating or ranking someone else's AI output is adjacent work, but it isn't this job.
• Enough technical ability to hold your own. You'll build some of your own tools and you'll be in conversations with engineers about integration and constraints. You don't need to ship production services, but you do need to figure things out quickly without becoming a bottleneck.
• Strong language and communication skills. A large part of this job is words - precision, tone, and explaining your reasoning to people who don't do what you do.
• Familiarity with LLM APIs (OpenAI, Anthropic, etc.) and conversational frameworks (e.g., LangChain).
• Excellent writing and linguistic skills, with attention to tone, clarity, and cultural sensitivity.
• Basic programming skills (Python preferred) for testing and prototyping prompts.
• Strong cross functional partnership and communications skills.

Nice-to-Have

• Background in healthtech, hardware, healthcare, caregiving, or other regulated industries.
• Multi-lingual proficiency.
• Experience with Voice Agent tools like VAPI