Principal Conversational & Agentic AI Architect

Verizon•Irving, TX

1d•Remote

About The Position

We are seeking a highly skilled Conversational & Agentic AI Architect to design, build, and optimize enterprise-grade conversational AI and multi-agent solutions within a Consulting organization. This role demands hands-on technical architecture, a product-led engineering mindset, and extensive expertise in both speech science and Generative AI. You will be responsible for consulting with customers to create seamless, multi-turn conversational experiences that move beyond rigid decision trees into dynamic, agentic interactions, successfully integrating complex backend systems while managing both sophisticated NLU models and LLMs.

Requirements

Bachelor’s degree or four or more years of work experience.
Six or more years of relevant experience required, demonstrated through one or a combination of work and/or military experience, or specialized training.
Proficiency in modern programming languages and frameworks such as Python, JavaScript, Java, Next JS, Node.js, React js
Strong working experience with GenAI, LLM Models, MCP, Vector DB, RAG, Vertex AI, Agentic AI frameworks like NGA, ADK or LangChain/LangGraph, creating AI agents.
Experience with Cloud platforms like GCP, Azure or AWS and cloud technologies like OpenStack, Terraform, Ansible or Chef
Experience working with LLM observability, analytics, evaluations, testing and annotation using tools like LangSmith, LangFuse, Streamlit, Arize or similar tools.
Experience working with AI/ML development.
Willingness to travel 25%

Nice To Haves

Background in Computational Linguistics or Data Science.
Certifications such as Microsoft Agentic AI Business Solutions Architect, Google Cloud Professional Architect, or Machine Learning Engineer.
Experience with Generative and NLU design.
Experience using Python libraries (Pandas/NumPy) for large-scale utterance data cleaning and optimization.
Familiarity with CI/CD pipelines for automated agent deployment and modern Product-Led operating models

Responsibilities

Architect and maintain complex conversational flows, transitioning from traditional State-Based Flows and event handlers to modern Generative AI and multi-agent orchestrated solutions (e.g., Dialogflow Playbooks, Vertex AI , or Microsoft Copilot Studio).
Manage complex intent taxonomies capable of handling 200+ unique intents without overlapping, ensuring high Intent Precision & Recall while minimizing False Positives.
Lead efforts in LLM fine-tuning, prompt engineering, distillation, and evaluating small/efficient models for latency-sensitive voice applications.
Conduct regular conflict detection and ambiguity testing to resolve intents that compete for the same user phrases, while securing AI models against prompt manipulation and data vulnerabilities.
Tune Speech-to-Text (STT) and Speech-to-Speech adaptations and hints to improve recognition for industry-specific jargon, and design complex custom entities to extract critical data from unstructured speech and generative interactions.
Build and deploy robust webhooks and serverless backend services using Node.js or Python to connect virtual agents with third-party APIs, CRMs, and databases.
Implement model serving infrastructure and scalable cross-platform AI solutions.
Build "Golden Sets" for automated regression testing to measure F1-scores across model versions, and use tools like BigQuery to analyze real-world "No-Match" logs.
Conduct Return-on-Investment (ROI) analysis of AI-powered solutions and interpret telemetry data to drive continuous product improvement.