Agentic AI Architect

ZENITH INFOTEK LLC•Research Triangle Park, NC

2d•$70 - $75•Hybrid

About The Position

The Customer is seeking an AI Architect to lead the technical design, orchestration, and implementation of their next-generation Agentic AI ecosystem. Reporting into the Customer Journey AI team, the architect will design blueprints for autonomous, goal-oriented AI agents to execute complex, end-to-end workflows across all customer touchpoints, including digital portals, the Spectrum TV App, conversational IVR, messaging, and retail/field operations. This role involves building a framework to transition customer experience from static chatbots to proactive, autonomous agents capable of real-time reasoning, tool use, and safe backend integration. The architect will bridge the gap between cutting-edge LLM orchestration and massive legacy telecom architectures (like BSS/OSS, CRM, and billing systems), focusing on multi-agent systems, real-time tool use, and safety guardrails to ensure autonomous agents can safely resolve customer issues without human intervention.

Requirements

10+ years in enterprise software architecture, with at least 3+ years specifically dedicated to production-grade AI/ML systems and LLM orchestration.
Proven track record of deploying systems utilizing multi-agent collaboration, autonomous reasoning (ReAct patterns), and complex function calling.
Advanced proficiency with Python, vector databases (e.g., Pinecone, Milvus, pgvector), and LLM frameworks.
Experience with open-source and commercial models (OpenAI, Anthropic, Llama, Mistral).
Deep experience with AWS or Google Cloud AI infrastructure, streaming data pipelines (Kafka, Flink), and caching strategies (Redis) for low-latency inference.
Familiarity with integrating AI layers on top of high-transaction enterprise architectures, REST/GraphQL APIs, and microservices.

Nice To Haves

Prior experience working with telecommunications customer data models, billing systems (Amdocs, CSG), or network operations.
Experience setting up automated evaluation rigs (e.g., Ragas, TruLens) to continuously score agent accuracy, hallucination rates, and task completion metrics in production.

Responsibilities

Design and implement a scalable, highly available architecture for multi-agent systems, utilizing state-of-the-art frameworks (e.g., LangGraph, AutoGen, CrewAI, or proprietary orchestration layers) to handle complex customer intents.
Architect robust, low-latency state-management systems that allow AI agents to maintain context across multi-session, multi-channel customer journeys.
Design safe, secure, and standardized interfaces enabling LLM agents to accurately call tools, query databases, and execute actions within legacy BSS/OSS, billing platforms, and network telemetry systems.
Build and maintain deterministic validation layers, evaluation pipelines, and guardrail frameworks (e.g., NeMo Guardrails, Llama Guard) to ensure agent behavior complies with strict telecom compliance, privacy laws, and brand guidelines.
Optimize Retrieval-Augmented Generation (RAG) pipelines and vector database infrastructure to give agents instantaneous access to thousands of internal knowledge bases, equipment manuals, and structured account schemas.
Collaborate closely with Software Development, Data Engineering, DevOps, and Product teams to transition legacy customer touchpoints into AI-native interfaces.