Senior Staff Software Engineer_Voice AI Agent Prompting

ServiceNow•Santa Clara, CA

9d•$190,900 - $334,100

About The Position

The Forward Deployed Engineer, Voice AI Agent Engineering designs, builds, ships, and operates production voice AI agents whose behavior is model-driven. These agents interpret caller intent over live audio, reason over context, invoke tools, and act on callers’ behalf. The role spans the full lifecycle: translating ambiguous call-flow and telephony requirements into precise specifications, authoring and maintaining agent instructions and prompts, and demonstrating through automated evaluation that agents behave correctly on real calls. It is not a machine-learning or AI-research role and does not involve training foundation or speech models. Instead, it combines software engineering with the design and evaluation of systems that select execution paths at runtime. The engineer treats natural-language instructions, prompts, tool descriptions, and guardrails as versioned, reviewable engineering artifacts. They address the demands of spoken interactions, including turn-taking, barge-in, and speech-recognition errors. Since behavior is probabilistic and occurs live, correctness is established through measured outcomes at scale, including latency, containment, and transcription accuracy. The engineer owns the correctness, quality, safety, and reliability of what reaches live callers, regardless of whether a human or an AI agent produced it. The role works directly within customers’ telephony, CTI, and CCaaS environments, building against their call data, integrations, and channels; tuning instructions and evaluations against live traffic; and bringing evidence back to improve the product. It also involves collaboration with product managers, designers, and platform teams to define success criteria, align on tradeoffs, and communicate capabilities and risks.

Requirements

Demonstrated track record building, shipping, and operating production voice applications, including delivery of AI-native voice-agent features used by real callers rather than demos.
Direct experience authoring agentic instructions and prompts for voice, designing AI-driven autonomous call workflows, and building evaluation and testing against live call traffic.
Willingness to work directly with customers in a forward-deployed capacity, embedded in their voice and contact-center environments.
At least 10 years of software-engineering experience, including at least 2 years deploying contemporary AI solutions at customers.
Bachelor’s degree in Computer Science, Software Engineering, or a related technical field, or equivalent practical experience.
Strong command of software-engineering fundamentals, including data structures, algorithms, system design, APIs, data modeling, and testing.
Hands-on experience building applications with model-driven behavior at their core, including application-layer agent orchestration, tool and function calling, context assembly, enterprise-data grounding, and management of live-audio latency, cost, and failure modes.
Demonstrated experience designing multi-step voice-agent workflows, including call-flow decomposition, decision points, autonomy boundaries, human-in-the-loop checkpoints, exception and retry handling, and observability.
Practical experience with telephony integration, speech recognition and synthesis, and streaming audio, including barge-in, turn-taking, and confirmation patterns.
Demonstrated skill in authoring and maintaining voice-agent instructions, tool descriptions, guardrails, refusal rules, and escalation rules with versioning, review, and regression coverage.
Voice-focused prompt-engineering skills, including task decomposition, golden and few-shot examples, structured outputs and schema enforcement, grounding and citation, and iteration based on call-level evaluation results.
Ability to build measurable evaluation frameworks for call quality, agent behavior, tool-selection accuracy, and regression risk, including datasets, scenario suites, model-as-judge scoring with human QA calibration, continuous evaluation, drift detection, and adversarial testing.
Ability to specify call-flow problems rigorously and exercise architectural judgment about whether to solve them in code, instructions, or through delegation to an AI agent.
Working knowledge of risks in AI-integrated voice applications, including prompt injection, sensitive-data and call-recording leakage, over-broad tool access, and unsafe autonomous actions, and ability to implement guardrails, least-privilege controls, and monitoring.
Comfort delivering directly with customers in their telephony environments under incomplete information and fixed timelines, exercising judgment about fast local solutions versus durable solutions, and communicating capabilities, limitations, and risks to non-engineers.

Nice To Haves

Advanced degrees or certifications; these do not substitute for a demonstrated record of shipping reliable production voice AI agents.
Prior forward-deployed, solution-engineering, or professional-services experience delivering software inside a customer’s environment.
Experience with telephony and contact-center platforms, IVR, speech recognition and synthesis, streaming audio, and real-time latency optimization.
Experience partnering with conversation or content designers on dialogue flow, tone, and error-recovery design.
Experience with evaluation frameworks, prompt and instruction management tools, LLM-application tracing, and large-scale analysis of production call transcripts.
Enterprise-domain depth in customer service, contact-center operations, sales, or enterprise workflow sufficient to challenge requirements rather than only implement them.

Responsibilities

Build and ship agentic voice applications involving intent interpretation, multi-step reasoning, tool invocation, and actions on callers’ behalf, together with the CTI, CCaaS, and telephony integrations required for production.
Design autonomous voice workflows by decomposing call flows and business processes, determining appropriate autonomy boundaries and human-agent handoffs, and defining handling for misrecognition, retries, and escalation.
Build voice experiences that preserve context across turns, handle barge-in and interruption, recover from speech-recognition errors, disambiguate caller intent, and confirm consequential actions within live-call latency constraints.
Write, structure, version, review, and maintain voice-agent system instructions, role definitions, tool descriptions, guardrails, and escalation rules, with regression coverage.
Iterate on prompts using measured call outcomes, including task decomposition, golden call transcripts, structured output schemas, grounding and citation, human hand-back, and token and latency costs.
Design and operate automated evaluation for non-deterministic voice behavior, including golden-call datasets, multi-turn suites, model-as-judge scoring calibrated to human QA, CI gates, drift detection, and adversarial, jailbreak, grounding, and tool-selection testing.
Translate customer call-flow and telephony requirements into testable specifications with explicit scope, constraints, and acceptance criteria; decompose work into agent-sized tasks and review AI coding-agent output for correctness and maintainability.
Work directly in customers’ telephony, CTI, and CCaaS environments, build against their data and integrations, tune instructions and evaluation sets against live traffic, and return evidence that can improve the product.
Monitor call containment, hallucinations, unsafe actions, and transcription accuracy; defend against prompt injection and data leakage; and incorporate production failures into specifications and evaluation sets.
Partner with product managers, designers, and platform teams to define success criteria, align on tradeoffs, and communicate capabilities and risks clearly.