Principal, System Engineer 2, Multi-Agent AI, AdTech

DIRECTV•New York, NY

6d•$178,952 - $268,307•Hybrid

About The Position

DIRECTV is seeking a hands-on senior individual-contributor systems engineer to design, build, and operationalize sophisticated agentic AI systems across its Advertising Technology ecosystem. The role provides end-to-end technical ownership for secure multi-agent systems and collaborative AI work environments, enabling business users and agents to plan, reason, use tools, share context, and complete complex advertising and media workflows. It combines architecture, production coding, platform services, enterprise integration, and direct engagement with business users. The position focuses on multi-agent architecture and development; collaborative workspaces and shared context; backend services, APIs, event-driven workflows, and governed enterprise integrations; context engineering for business workflows; production evaluation, observability, safety, and governance; and technical leadership and forward deployment. The engineer partners with Product, Ad Sales, Operations, Data, and Engineering to turn ambiguous needs into deployable solutions, while balancing experimentation with maintainability, security, governance, and measurable business value. The position is based in New York City and requires in-office attendance two days per week. Approximately 10% travel may be required. Company-paid relocation assistance is not available. A background check may be required because of routine access to proprietary company and customer data. The listed base salary range is $178,952–$268,307 for wage zone N4; compensation depends on work location and other job-related factors. Bonus and benefits are not included in the salary range.

Requirements

Bachelor's or Master's degree in Computer Science, Engineering, or a related technical field.
5–7 years of progressive software, systems, platform, or architecture experience, including hands-on ownership of production services.
Proven experience building and deploying agentic AI or GenAI systems beyond proofs of concept.
Demonstrated experience leading solutions from discovery through production and making end-to-end technical decisions.
Strong coding experience in Python and proficiency in at least one additional modern language such as TypeScript, Java, Go, or C#.
Deep knowledge of LLM applications, multi-agent orchestration, tool use, structured outputs, memory, RAG, context engineering, evaluation, and human-in-the-loop controls.
Strong background in backend engineering, API design, microservices, distributed systems, event-driven architecture, and cloud-native deployment.
Hands-on experience with frameworks and platforms such as LangGraph, Semantic Kernel, Strands, Bedrock AgentCore, Snowflake Cortex, or comparable technologies.
Experience implementing secure enterprise integrations, identity and authorization, observability, automated testing, and CI/CD.
Ability to explain complex system and agent architectures, integration decisions, risks, and tradeoffs to technical and non-technical audiences.
Proven ability to lead through influence, write and review production code, mentor engineers, and drive decisions across teams.
Product mindset and judgment to simplify designs and focus teams on measurable business outcomes.
Ability to meet the hybrid work requirement of in-office attendance two days per week in New York City.

Nice To Haves

Hands-on experience in advertising, media, AdTech, MarTech, campaign management, ad operations, programmatic advertising, measurement, or monetization.
Experience building cowork, collaborative agent, case-management, investigation, or human-agent work environments.
Experience across application engineering, platform engineering, data engineering, ML engineering, product development, SRE, or solution architecture.
Experience with AWS, Kubernetes, serverless services, Snowflake, vector or graph stores, and enterprise data platforms.
Open-source contributions, technical publications, patents, or demonstrated thought leadership in agentic systems.

Responsibilities

Design and build production multi-agent systems, including planners, coordinators, specialist agents, reviewers, human approval points, shared state, memory, and recovery paths.
Define orchestration patterns for agent delegation, task decomposition, consensus, tool selection, and deterministic workflow control.
Write production code and reference implementations.
Evaluate model, framework, and runtime tradeoffs while keeping solutions modular and avoiding unnecessary vendor lock-in.
Build collaborative workspaces where business users and multiple agents can jointly investigate, plan, create, review, and execute work.
Design shared-context patterns such as workspaces, case files, task state, artifacts, approvals, audit history, and role-aware memory.
Make agent evidence, tool activity, decisions, and handoffs understandable to business users.
Enable reusable skills and workflows for campaign planning, operations, analytics, incident response, and engineering use cases.
Design and develop backend services, APIs, event-driven workflows, tool registries, and integration services connecting agents to enterprise capabilities.
Build governed tool interfaces using MCP-style patterns, schemas, identity propagation, authorization, validation, rate limits, and auditability.
Integrate agent systems with advertising platforms, data services, Snowflake, operational systems, content repositories, and third-party APIs.
Apply distributed-systems practices for reliability, idempotency, retries, timeouts, fallbacks, concurrency, and state management.
Design context strategies combining user intent, role, workflow state, enterprise knowledge, structured data, tool outputs, conversation history, and policy constraints.
Build retrieval and context pipelines that deliver relevant information to agents while controlling noise, latency, cost, and sensitive-data exposure.
Develop reusable instruction, skill, memory, and context templates for business personas and workflows.
Partner with business users to understand their work, encode domain knowledge, and improve agent behavior through feedback and evaluation.
Establish evaluation strategies for task success, groundedness, tool correctness, trajectory quality, safety, latency, reliability, and cost.
Implement observability across agent steps, model calls, context assembly, tool activity, decisions, and business outcomes.
Design guardrails for unsafe actions, prompt injection, data leakage, runaway loops, failed tools, and low-confidence decisions.
Drive production readiness through automated testing, CI/CD, release controls, incident response, runbooks, and continuous improvement.
Work with Product, Ad Sales, Operations, Data, and Engineering to turn ambiguous problems into deployable agentic solutions.
Lead architecture and code reviews, mentor engineers, establish reusable standards, and manage cross-system dependencies and risks.
Balance rapid experimentation with maintainability, security, governance, and measurable business value.