Prompt Engineer – Voice AI Agents

At SPsoft, we build Voice AI solutions for customer communication. We are looking for a Prompt Engineer to create, test, and optimize prompts for Voice AI agents.

This role is focused on prompt writing, conversation design, testing, and improving LLM responses. You will create and refine the instructions that define how our AI agents communicate with users, handle different scenarios, follow business rules, use the available information, and respond naturally during voice conversations.

This is not a software development position. Your primary tools will be prompts, conversation flows, test scenarios, and LLMs, not software code. You should, however, understand how prompts operate within a production Voice AI system and be able to work closely with engineers when an issue involves integrations, speech recognition, speech synthesis, or application logic.

Key Responsibilities

• Design, write, and maintain system prompts, conversation instructions, behavior rules, guardrails, and reusable prompt components for Voice AI agents.
• Define how AI agents greet users, identify intent, ask clarifying questions, handle ambiguity, manage interruptions, respond to unexpected situations, escalate conversations, and close calls.
• Write prompts specifically for voice interactions, ensuring that responses are concise, natural, easy to understand, and appropriate for spoken conversations.
• Adapt prompts to different business use cases, user roles, communication styles, and conversation scenarios.
• Test prompts through realistic multi-turn conversations, edge cases, failure scenarios, and alternative user responses.
• Evaluate whether AI-agent responses are accurate, relevant, consistent, natural, and aligned with the required business rules.
• Identify recurring failure patterns and determine whether they are caused by prompts, conversation logic, available context, speech recognition, speech synthesis, integrations, or model behavior.
• Run structured experiments with prompt wording, context, examples, instructions, tone, and response constraints.
• Create test scenarios, evaluation criteria, scorecards, and regression checks for prompt changes.
• Review approved conversation data and test results to identify opportunities for improvement.
• Work with engineers to implement and test prompts within production Voice AI workflows.
• Maintain prompt versions, documentation, test cases, change history, and reusable prompt libraries.
• Validate every material prompt change before it reaches production.

Required Skills and Qualifications

• Practical experience working with large language models and writing prompts for conversational AI.
• Strong understanding of system prompts, context management, instruction hierarchy, guardrails, examples, output constraints, and common LLM failure modes.
• Experience designing multi-turn conversations rather than only writing isolated prompts or single-response instructions.
• Ability to translate business requirements into clear AI-agent behavior, conversation rules, and testable scenarios.
• Strong written English and the ability to write precise instructions and natural spoken dialogue.
• Experience testing LLM responses systematically and improving prompts based on observed results.
• Strong analytical and debugging skills, including the ability to identify why an AI agent produced an incorrect or inconsistent response.
• Familiarity with Voice AI, speech-to-text, text-to-speech, latency, interruptions, and other constraints of real-time voice conversations.
• Ability to work closely with product managers, domain experts, QA specialists, and software engineers.
• Strong attention to detail and clear ownership of prompt quality and AI-agent behavior.

Will Be a Plus

• Experience creating evaluation datasets, automated prompt tests, or regression-testing workflows.
• Basic knowledge of Python, JSON, APIs, or other technical tools used to test and manage prompts.
• Experience working in regulated environments with HIPAA, SOC 2, or similar requirements.
• Experience with healthcare, telemedicine, customer support, sales, scheduling, or other conversational AI use cases.
• Experience with speech synthesis, prosody, pronunciation rules, or emotion-aware voice technologies.
• Experience designing prompts for multiple languages or regional language variations.

What’s in It for You

• Work directly on production Voice AI solutions and see how prompt changes affect real conversations.
• Own how AI agents communicate, respond, and behave across different scenarios.
• Develop quickly through hands-on experimentation, testing, and collaboration with product and engineering teams.
• Help establish prompt-engineering and conversation-design practices within a growing Voice AI team.