New opportunity

Senior Machine Learning Engineer- Generative AI (REMOTE)

The Home Depot · Atlanta, United States

About this role

The Senior Machine Learning Engineer designs, builds, integrates, optimizes, and maintains AI-powered applications using generative models. The role works independently and with product, UX, engineering, infrastructure, and business teams to deliver and support production AI products.

Responsibilities

Design and implement applications using large language models and other generative models, potentially including prompt engineering, model integration, Retrieval-Augmented Generation pipelines, and scalable AI services. Evaluate, test, monitor, and optimize AI systems; improve prompts and workflows; and create documentation or enablement materials.
Collaborate on secure, reliable, scalable solutions and developer-ready user stories. Review code and prompt implementations against engineering and responsible AI practices. Document work and meet quality and change-control standards.
Automate infrastructure, monitoring, and tests, including destructive testing for production resilience. Configure commercial software to meet business needs and build dashboards, logging, and alerts.
Support production applications and other product teams. Monitor service-level objectives, performance, capacity, and prediction quality. Participate in ongoing learning about software design and machine learning.

Requirements and preferences:

Minimum qualifications are age 18 or older, legal permission to work in the United States, a high school diploma and/or GED or equivalent knowledge and skills, and 2 years of work experience. No leadership experience or certification is required.
Preferred qualifications include 3–5 years of relevant experience; Python; building LLM applications; prompt optimization and evaluation; model APIs from Google, OpenAI, or Anthropic; Google Agent Development Kit; RAG and vector databases such as Vertex AI Search; and conversational AI or assistants. Other preferences include responsible AI, graph databases, knowledge ingestion, data mesh, CI/CD, production AI evaluation and monitoring, Google Cloud and its AI/ML services, big-data platforms, GPU acceleration, front-end development, SQL, resilient production systems, REST services, A/B testing, and advanced machine learning techniques.

Location and compensation: This is a remote/virtual role listed in Atlanta, GA. Remote roles may require living within commuting distance of an assigned office for occasional in-person work. Overnight travel is typically 5%–20%. The stated pay range is $100,000.00–$180,000.00; starting pay varies by position, location, education, training, and experience. Additional benefits may include a performance-based bonus or profit sharing, depending on the position.

For full details about this role, click “Apply on company site” to view the employer’s original posting.

Skills for this role

Generative AIMachine LearningLarge Language ModelsPythonPrompt EngineeringPrompt OptimizationPrompt EvaluationRetrieval-Augmented GenerationVector DatabasesVertex AI SearchGoogle Agent Development KitConversational AIAI AssistantsModel EvaluationProduction MonitoringResponsible AIBias MitigationSafety GuardrailsGraph DatabasesKnowledge IngestionData MeshCI/CDGoogle Cloud PlatformVertex AIBigQuery MLBigQueryData EngineeringCUDAcuDNNNode.js

Checking your profile…

Preparation suggestions, not the employer’s actual interview questions. Always check the original posting for current requirements.

Understand the role

What matters most

Deliver production generative AI applications
The role designs, builds, integrates, and maintains AI-powered applications; the posting names LLM integration and possible RAG pipelines. Prepare to explain design decisions across an application’s lifecycle, from user need to production support.
Turn cross-functional needs into testable delivery
Delivery and execution account for 70% of the listed responsibilities. The engineer works with product, UX, and engineering to create secure, reliable, scalable solutions and developer-ready user stories. Prepare an example of resolving ambiguity and defining acceptance criteria.
Evaluate quality and responsible use
The posting calls for AI evaluation, testing, prompt and workflow improvement, and possible reviews of code and prompts against responsible AI practices. Prepare to describe how evaluation findings would guide changes without overlooking safety.
Build resilience through automation
Duties include automating infrastructure, monitoring, and tests; running destructive tests; and meeting quality and change-control standards. Prepare to explain what a resilience test reveals and how a change can be released safely.
Support and observe production systems
Support and enablement account for 20% of the listed responsibilities. The role monitors service-level objectives, capacity, performance, and prediction quality, and supports other teams. Prepare to connect dashboards, logs, and alerts to diagnosis and follow-up action.

Likely challenges to think through

Balancing answer quality with system performance
A likely challenge, not a reported problem: prompt or retrieval changes could improve answer quality while affecting latency or capacity. Prepare a way to compare both quality and operational measures before and after a change.
Testing failures without disrupting delivery
A likely challenge, not a reported problem: destructive testing can reveal weaknesses but must be reconciled with production reliability and change controls. Prepare to discuss test scope, safeguards, and how findings would be prioritized.
Aligning diverse stakeholders on AI behavior
A likely challenge, not a reported problem: business expectations for an AI feature may be hard to express as testable engineering work. Prepare an approach for clarifying intended behavior, boundaries, and acceptance criteria with product and technical partners.

Practice your answers

Technical practice

How would you design an LLM-powered application for production?

What to coverSuggested preparation: Explain how you would choose and integrate a model API, structure the application service, and handle failures. Discuss trade-offs among response quality, latency, scalability, security, and maintainability.

How would you build and troubleshoot a RAG pipeline?

What to coverSuggested preparation: Walk through knowledge ingestion, embeddings, retrieval from a vector database, and grounding the generated answer. Explain how you would distinguish retrieval failures from generation failures and evaluate whether changes improve answers.

How would you optimize and evaluate prompts?

What to coverSuggested preparation: Describe a repeatable test set and criteria for answer quality, safety, and consistency. Explain how you would compare prompt or workflow changes without relying only on a few persuasive examples.

How would you design a conversational AI assistant?

What to coverSuggested preparation: Discuss handling conversation context, unclear requests, and model or tool failures. The posting prefers experience with assistants and Google Agent Development Kit; if relevant, explain when an agent framework helps and when a simpler design is preferable.

How would you evaluate a production AI change?

What to coverSuggested preparation: Compare offline evaluation with production monitoring and, where appropriate, A/B testing. Address both prediction quality and operational effects such as latency or capacity before deciding whether to retain a change.

How would you apply responsible AI practices to an LLM application?

What to coverSuggested preparation: Explain how you would identify bias or unsafe outputs, apply guardrails, and test their effectiveness. Discuss the trade-off between restricting harmful behavior and preserving useful responses.

When would you use graph data alongside vector search?

What to coverSuggested preparation: The posting lists graph databases, knowledge ingestion, and data mesh as preferences. Compare relationship-based queries with semantic retrieval, and explain how you would assess data freshness and quality before using either in an AI knowledge system.

How would you monitor and diagnose a production AI service?

What to coverSuggested preparation: Describe service-level objectives, dashboards, logs, and alerts covering performance, capacity, and prediction quality. Explain how you would isolate a problem in code, infrastructure, data, message processing, or model behavior.

How would you test an AI service for resilience?

What to coverSuggested preparation: The role calls for automated tests and destructive testing. Describe failures you would simulate, how you would verify recovery and high availability, and the safeguards you would use when testing a production system.

How would you deploy and change an AI application safely?

What to coverSuggested preparation: Explain how CI/CD, automated tests, code and prompt review, documentation, and change control can work together. Address how you would detect a regression and respond after release.

Behavioural practice

Tell me about turning an ambiguous request into testable work.

What to coverSuggested preparation: Use a real situation involving product, UX, engineering, or business partners. Describe how you clarified the need, shaped developer-ready work, and what resulted.

Describe a time you improved someone else's implementation.

What to coverSuggested preparation: Give a real example of reviewing code, an AI workflow, or another technical design. Explain the feedback you gave, how you handled disagreement, and the resulting improvement.

Tell me about a production issue you helped resolve.

What to coverSuggested preparation: Set out the situation, your diagnostic and communication actions, and the result. Include how you supported affected teams and what you changed to improve monitoring or resilience.

Describe a time you delivered with limited guidance.

What to coverSuggested preparation: Use a real example showing how you made decisions while keeping relevant partners informed. Explain the trade-offs you made and the outcome, rather than presenting independent work as work done in isolation.

Tell me about learning a new technique and applying it.

What to coverSuggested preparation: Describe the situation, how you evaluated what to learn, and the action you took. Explain the result and how you shared useful knowledge or documentation with others.

Sharpen your knowledge

Topics to prepare

LLM application design
Posting: The role designs and integrates generative-model applications and scalable AI services; model APIs from Google, OpenAI, or Anthropic are preferred experience. Suggested exercise: sketch an assistant architecture and explain model integration, latency, security, and failure-handling trade-offs.
Retrieval-Augmented Generation
Posting: RAG pipelines, vector databases, knowledge ingestion, and Vertex AI Search appear in the role description or preferences. Suggested exercise: trace a document from ingestion through retrieval to an answer, identifying where stale or irrelevant evidence could enter.
AI evaluation and responsible AI
Posting: The role may evaluate and optimize production AI systems and review prompts against responsible AI practices; bias mitigation and safety guardrails are preferred experience. Suggested exercise: define quality and safety checks for a conversational assistant, including how to review failed responses.
Production reliability and service objectives
Posting: The role monitors service-level objectives, capacity, performance, and prediction quality, and uses destructive testing to check resilience. Suggested exercise: work through an AI-service outage scenario and explain what to measure, alert on, and test before restoring service.
Cross-functional delivery and change control
Posting: The role works with product, UX, engineering, and infrastructure teams on testable user stories, documentation, reviews, and quality standards. Suggested exercise: turn a proposed AI feature into an acceptance-tested user story and identify what a code and prompt review should cover.

Skills to revise

Integrate a model API in Python
Posting: Python and model-API integration are preferred experience. Suggested exercise: write a small Python service that sends a request, handles errors and timeouts, and records enough information to investigate a failed response.
Build and test a RAG workflow
Posting: Implementing RAG with vector databases is preferred experience. Suggested exercise: create a small ingestion-and-retrieval workflow using sample documents, then test whether answers are supported by retrieved material; Vertex AI Search is one named option, not a requirement for the exercise.
Iterate on prompts with an evaluation set
Posting: Prompt engineering, optimization, evaluation, and testing are named activities. Suggested exercise: assemble representative requests, compare two prompt versions against defined quality and safety criteria, and document the observed trade-offs.
Instrument and stress-test an AI service
Posting: The role creates dashboards, logging, alerts, automated tests, and destructive tests for production resilience. Suggested exercise: add request logging and failure alerts to a sample service, then simulate a dependency failure and check whether the service and alerts behave as intended.
Prepare a reviewed deployment change
Posting: CI/CD is preferred experience, while documentation, code review, and change-control standards are responsibilities. Suggested exercise: draft an automated test and deployment checklist for a prompt or model-integration change, including rollback criteria and a concise review note.

Your preparation checklist

Map your evidence to the role
Prepare a one-page list of examples relevant to the posting’s production GenAI work and preferred skills, such as Python, LLM applications, RAG, model APIs, and Google Cloud. Mark which examples are direct experience and which topics you would discuss as design approaches.
Rehearse a production AI project walkthrough
Choose a real project, if available, and explain the problem, your contribution, architecture, tradeoffs, testing, deployment, and results. Be ready to distinguish your decisions from team decisions.
Practice an LLM application design exercise
As interview practice, sketch a conversational application with a model API and optional RAG pipeline. Show ingestion, retrieval, service boundaries, security, scaling choices, and how you would handle failures; do not assume this is a planned Home Depot project.
Prepare an evaluation and safety plan
Draft a compact example of prompt and response evaluation: test cases, quality measures, regression checks, bias and safety checks, and a method for comparing changes. The posting calls for evaluation, optimization, and responsible AI review.
Show how you would operate the system
Prepare a sample production checklist covering CI/CD, automated tests, resilience or destructive testing, service-level objectives, prediction quality, capacity, dashboards, logs, alerts, and incident response.
Rehearse a cross-functional delivery example
Prepare a concise account of turning a product need into a testable developer-ready story, collaborating with product, UX, and engineering, reviewing code or prompts, and documenting a change. Use a real example where possible.
Write focused questions for the employer
Ask which GenAI applications this role supports, how the team measures production quality and service-level objectives, how responsible AI reviews work, and how delivery and support responsibilities are shared. Clarify remote in-person expectations and travel if relevant.

Availability last checked: Nov 4, 2026. Status can change after a check; confirm on the employer’s website. Report an issue · Our checking policy

KEEP EXPLORING

Similar AI jobs

Related skills and specializations in United States. Matched to this role, not your profile.

Explore more jobs

Finding similar opportunities…