Senior AI Engineer – Forward Deployed, Agent Platform (REMOTE)
At a glance
Senior Solutions & Sales Engineering role at Homedepot. Remote, United States · full-time · $100,000–$180,000 base.
growthroles summary, based on the employer's posting.
What you'll do
- Work with business teams to map operational workflows, set boundaries for agent behavior and agree on success checks before development
- Build and launch multi-step agents in Python, FastAPI, Google ADK and Gemini Enterprise, including approval steps for actions that change data
- Create secure MCP servers and REST connections to enterprise systems, then publish reusable integrations for other teams
- Develop evaluation datasets and regression checks, and turn successful builds into templates that reduce later implementation effort
- Review agent submissions, prepare operating guides and training, and hand off ongoing ownership to business teams
What you bring
- At least six years building and running production Python services, including FastAPI, asynchronous services or REST APIs
- Hands-on experience orchestrating multi-step agents and working with LLM APIs such as Gemini, Claude or OpenAI
- Experience integrating enterprise SaaS with MCP and managing access through controls such as OAuth, SSO or role-based permissions
- Practical experience with agent evaluation, golden datasets and resilient production engineering
- You must be legally authorized to work in the United States
Who this fits
This senior role suits an engineer who can work directly with nontechnical business partners, navigate unclear workflows and explain technical tradeoffs to executives. It has no direct reports and includes overnight travel, typically five to twenty percent of the time. The position is remote in the United States.
From the employer
With a career at The Home Depot, you can be yourself and also be part of something bigger.
Position Purpose:
- Embed with THD Business Verticals: Partner directly with Finance, Marketing, Supply Chain, Online, Store Ops, HR, and Legal/Risk, etc., to turn ambiguous operational processes into production-ready, enterprise-governed AI agents.
- Educate & Enable: Teach verticals what surfaces they can deploy on (Gemini Enterprise, Google ADK, Skills, Agent Designer, MCP servers, Canvas, Agent Gateway/Registry) and how to manage and evolve their agents. Consult business on technical approaches to build agents.
- Drive Adoption & Reuse: Grow Gemini Enterprise adoption across the enterprise – team agents, individual Sparks, and a marketplace of reusable enterprise agents, Skills and tools. Treat every engagement as supply for the marketplace and actively promote reuse so teams compose existing assets before building new ones.
- Explicit Scope Guardrail: Foundation models are consumed as managed services—no pre-training/fine-tuning models, no GPU/CUDA kernel work, and no classical feature stores or model registries.
Responsibilities:
- Embed & Discover Workflows: Sit with non-technical business partners to elicit requirements, define negative scope ("what the agent must never do"), and lock in acceptance criteria before coding.
- Build & Deploy Production Agents: Ship end-to-end multi-step agents and conversational assistants in Python (FastAPI), Google ADK, and Gemini Enterprise, including Human-in-the-Loop (HITL) confirmation gates for write actions.
- Engineer Enterprise Integrations & MCP Servers: Build resilient MCP servers and REST integrations for enterprise systems (e.g. Jira, Confluence, SharePoint/M365, Slack, ServiceNow, Salesforce, etc.) with strict OAuth/IAM boundaries, idempotency, retries, and High Availability patterns; publish them as reusable assets in the enterprise marketplace.
- Establish Reusable Evaluation Harnesses: Build workflow-level golden datasets, prompt/agent evaluations, and automated regression suites so business owners have an objective pass/fail gate.
- Templatize & Build the Paved Road: Extract reusable templates, Skills, and packaging standards from every build so the second vertical build takes ≤60% of the effort of the first.
- Review Submissions & Govern Quality: Review partner-submitted agent code, company Skills, and prompt implementations for the Agent Sharing & Registry workflow against engineering, security, and Responsible AI standards;curate the marketplace so published agents, Skills and tools are discoverable, documented and trusted.
- Execute Ownership Handoff & Enablement: Author runbooks and enablement materials, verify receiving-owner readiness, and cleanly transfer day-to-day operations to the vertical.
- Feed Platform Architecture: Convert recurring vertical needs into scoped backlog items for the core APEX platform (Gateway/Registry, Model Armor guardrails, observability, and lifecycle management).
- Enable Builders & Grow Adoption: Support product to run technical workshops, office hours; review low-code agents built by business teams; maintain documentation and a pattern library; and use usage data and feedback to improve adoption and the platform.
Key Responsibilities:
- 70% Delivery and Execution - Collaborates and pairs with other product team members (UX, engineering, and product management) to create secure, reliable, scalable machine learning solutions; Documents, reviews, and ensures that all quality and change control standards are met; Works with Product Team to ensure user stories that are developer-ready, easy to understand, and testable; Writes custom code or scripts to automate infrastructure, monitoring services, and test cases; Writes custom code or scripts to do "destructive testing" to ensure adequate resiliency in production; Configures commercial off the shelf solutions to align with evolving business needs; Creates meaningful dashboards, logging, alerting, and responses to ensure that issues are captured and addressed proactively
- 10% Learning - Participates in learning activities around modern software design, machine learning, and development core practices (communities of practice); Proactively views articles, tutorials, and videos to learn about new technologies and best practices being used within other technology organizations
- 20% Support and Enablement - Fields questions from other product teams or support teams; Monitors tools and participates in conversations to encourage collaboration across product teams; Provides application support for software running in production; Proactively monitors production Service Level Objectives for products; Proactively reviews the Performance and Capacity of all aspects of production: code, infrastructure, data, message processing, and prediction quality
Direct Manager/Direct Reports:
- This Position typically reports to Software Engineer Manager or Sr. Software Engineer Manager
- This Position has 0 Direct Reports
Travel Requirements:
- Typically requires overnight travel 5% to 20% of the time.
Physical Requirements:
- Most of the time is spent sitting in a comfortable position and there is frequent opportunity to move about. On rare occasions there may be a need to move or lift light articles.
Working Conditions:
- Located in a comfortable indoor area. Any unpleasant conditions would be infrequent and not objectionable.
Minimum Qualifications:
- Must be eighteen years of age or older.
- Must be legally permitted to work in the United States.
Preferred Qualifications:
- 6+ yrs shipping/operating production Python services (FastAPI, async microservices, REST)
- Multi-step agent orchestration (Google ADK, LangGraph, CrewAI, AutoGen, or equivalent) & LLM APIs (Gemini, Claude, OpenAI)
- Systematic prompt engineering, structured outputs, context optimization & evaluation
- Enterprise SaaS & MCP integration + auth boundaries (OAuth2, SSO, service accounts, delegated user, RBAC)
- RAG, knowledge ingestion & search (Vertex AI Search / Gemini Enterprise Data Stores, access-scoped retrieval)
- Agent evaluation harnesses, golden datasets & HA/defensive coding practices
- GCP (Vertex AI, Cloud Run, BigQuery), analytical SQL, CI/CD, and observability (Grafana, Prometheus, OpenTelemetry)
- Forward-deployed execution: cross-functional collaboration across skill levels, code/prompt review, enablement docs & handoff
- Customer-facing delivery & communication: demonstrated experience working with nontechnical stakeholders, comfort with ambiguity, and presenting technical tradeoffs to executives (assessed in the interview via a mock discovery session). ● Security & Responsible AI fundamentals for agents that take write actions: PII handling, HITL approvals, audit trails, and least-privilege access.
- Direct hands-on experience with Gemini Enterprise, Vertex AI Search / Data Stores, Google ADK, and MCP / FastMCP
- Prior Forward Deployed, Solutions, or Embedded Platform Engineer experience
- HITL approval workflows, write-action safety controls & immutable audit trails
- Responsible AI & safety guardrails in prod (Model Armor, prompt-injection defense, bias/hallucination mitigation, PII redaction)
- A/B testing, graph/connected knowledge systems, Terraform/HCL, Docker/GKE, and TypeScript/React/JS for UI/Canvas surfaces
- Retail, eCommerce, supply chain, store ops, or finance domain background
Minimum Education:
- The knowledge, skills and abilities typically acquired through the completion of a high school diploma and/or GED.
Preferred Education:
- No additional education
Minimum Years of Work Experience:
- 2
Preferred Years of Work Experience:
- No additional years of experience
Minimum Leadership Experience:
- None
Preferred Leadership Experience:
- None
Certifications:
- None
Competencies:
- Global Perspective
- Manages Ambiguity
- Nimble Learning
- Self-Development
- Collaborates
- Cultivates Innovation
- Situational Adaptability
- Communicates Effectively
- Drives Results
- Interpersonal Savvy
Benefits offered include health care benefits, 401K, ESPP, paid time off, and success sharing bonus. For a full list of the various benefits The Home Depot offers, visit https://careers.homedepot.com/our-benefits.
For California, Colorado, Connecticut, Rhode Island, Nevada, New York City, Ithaca (NY), Westchester County (NY), and Washington residents: