Senior AI Architect

Nexstar Media Group, Inc.•New York, NY

2d•Onsite

About The Position

Nexstar is seeking a senior AI Architect to lead our efforts to design, develop and deploy AI and ML models and agents on top of a new single governed data estate to drive greater efficiency and effectiveness across the business. This role is responsible for guiding design, development and deployment of data architecture to enable efficient and effective AI modeling and agentic development across a large, multi-phase data platform. This means designing the systems, patterns, and standards that enable agents to answer business questions accurately, safely, and at scale across a growing set of data domains and use cases.

Requirements

Deep experience designing and deploying production AI agent systems, including LLM selection, orchestration, and evaluation
Familiarity with Databricks platform and related products
Strong understanding of how agents interact with structured data, including retrieval patterns, semantic layers, and metadata-driven context
Familiarity with major LLM platforms and orchestration frameworks, with the ability to evaluate tradeoffs objectively across vendors
Experience designing safety and governance patterns for AI systems
Comfortable working across data engineering, analytics, and business stakeholder teams

Nice To Haves

Experience in media, advertising, or other data-intensive industries with complex business metrics and multi-system data environments helpful

Responsibilities

Define the overall agent architecture: how agents are structured, how they access data, how they are governed, and how the architecture scales as new data domains and use cases come online
Design and own the central agent integration layer that mediates all data access between agents and the platform, so that guardrails and controls are enforced in one place and consistently applied across all agents built on the architecture
Evaluate and select LLM frameworks and orchestration approaches, including Databricks Mosaic AI and other platforms, matching technology choices to use case requirements rather than defaulting to a single stack
Establish the patterns governing how agents interact with governed data, ensuring agents operate within access controls, return accurate and trustworthy answers, and handle business context correctly
Define guardrails across the agent layer: read-only data access, user- and role-based data access controls, PII handling, output validation, and hallucination mitigation
Design the agent evaluation framework, including synthetic query suites, accuracy validation, and the use of agents to generate and run test variations against domain coverage and answer consistency, establishing the criteria that determine when an agent is production-ready
Define user-facing agent experience patterns: conversational vs single-query, output formats, how agents surface caveats and supporting data, and how they are accessed by different user types
Work with data and analytics teams to ensure the underlying data layer (schemas, metadata, metric definitions) is structured to support accurate agent responses
Ensure agent architecture respects multi-tenant data boundaries as the platform scales to support additional broadcast partners
Stay current on the rapidly evolving AI and agent tooling landscape and bring relevant developments into the architecture proactively
Guide and set standards for the AI developers building agents across the platform, ensuring individual agent work builds consistently on the central architecture.