Lead Data Architect

Domino's•Ann Arbor, MI

14h

About The Position

The posting title is Lead Data Architect, while the description identifies the role as Lead AI Architect. This is a senior, hands-on individual-contributor role serving as the technical authority for Domino’s enterprise AI platform across classical, generative, and agentic AI. The architect owns the AI reference architecture, roadmap, and paved road for development, serving, evaluation, integration, and governance. The role exercises design authority and influences teams without direct reports, making explicit, evidence-based decisions about what to build for production, prototype, scaffold for later, or defer. The role requires enterprise-scale AI platform delivery experience and continued hands-on work, including prototypes and reference implementations. It spans semantic search and retrieval-augmented generation, model and prompt management, multi-model routing, agent orchestration, AI governance and security, observability and economics, and Azure-native platform engineering. The environment is Azure-primary with Databricks. The architect partners across engineering, architecture, data governance, security, legal, privacy, risk, finance, and business teams to turn strategy into governed, measurable AI services and production outcomes.

Requirements

Deep enterprise architecture and systems-design expertise, including end-to-end AI platforms across applications, models, data, APIs, integration, identity, infrastructure, and operations.
Strong grounding in distributed systems, event-driven and service-oriented patterns, resilience, high availability, fault tolerance, and disciplined architectural decision-making.
Deep understanding of classical machine learning, deep learning, NLP, foundation models, transformers, embeddings, semantic retrieval, RAG, fine-tuning, multimodal AI, and agentic or multi-agent architectures, including their trade-offs and failure modes.
Ability to design structured and unstructured data ingestion and transformation, chunking and indexing, vector and hybrid retrieval, knowledge graphs, metadata models, lineage, provenance, data quality, and access controls.
Ability to ground conversational experiences in structured data, including text-to-SQL and multi-turn dialogue.
Experience establishing evaluation frameworks and benchmarks, including recall@k, NDCG, groundedness, and LLM-as-judge; automated testing, release controls, model and prompt versioning, observability, quality monitoring, incident response, rollback, and continual improvement using MLOps and LLMOps principles.
Ability to build identity, authorization, privacy, model security, human oversight, regulatory controls, and auditability into architecture from the start; understanding of prompt injection, sensitive-data leakage, unsafe agent actions, and model or software supply-chain risk.
Ability to make quantitative architecture decisions involving latency, throughput, concurrency, availability, capacity, inference architecture, compute utilization, quality/performance trade-offs, consumption forecasting, and cost optimization, including AI FinOps at scale.
Ability to translate business strategy into qualified AI use cases, measurable outcomes, target-state architectures, investment priorities, and delivery roadmaps, and to connect solutions to operational or financial KPIs.
Demonstrated ability to deliver from discovery and architecture through experimentation, production deployment, and operational improvement.
Ability to build or review prototypes using Python, SQL, and Spark, validate assumptions, review implementation designs, diagnose cross-layer failures, and defend technical trade-offs.
Production depth in DevSecOps, Infrastructure as Code, Kubernetes, and model serving.
Experience defining reusable patterns and standards, leading architecture reviews, mentoring engineers and architects, and aligning product, data science, engineering, infrastructure, security, and business teams.
Ability to create clarity amid ambiguity and communicate effectively with executives and practitioners.
A strong record of leading complex AI or ML solutions in production at enterprise scale; typically 10+ years across software, data, or ML engineering with significant architecture leadership.

Nice To Haves

Hands-on experience with Microsoft technologies including Azure AI Foundry and Agent Service, Azure OpenAI, Azure ML, AI Search, API Management, Functions, Databricks, and Copilot Studio.
Experience with agent frameworks such as LangGraph or Semantic Kernel and MCP.
Experience in retail, restaurant, e-commerce, or supply chain, including high-volume real-time workloads, streaming technologies such as Kafka or Event Hubs, and model risk management.
Experience with production multi-agent systems and agent runtimes, policy-aware agent gateways with authenticated tool routing and server registries, and GPU-accelerated serving stacks.
Contributions to open-source AI projects, published work, or recognized community involvement.

Responsibilities

Own the enterprise AI reference architecture and establish scalable standards, principles, and design patterns across business groups.
Own the roadmap for agentic systems platforms, translating business objectives into sequenced technical strategy, milestones, and reasoned build-depth decisions for each use case.
Design and maintain the paved road for AI development, including patterns for building, serving, evaluating, and integrating models and agents.
Guide AI compute and token-usage optimization; lead evidence-based build-versus-buy, model, and framework decisions using benchmarks, cost per answer, and real-load latency.
Design chunking, embedding, indexing, query construction, hybrid retrieval, fusion, re-ranking, and re-embedding/versioning strategies, measuring retrieval quality and managing precision, recall, and latency trade-offs.
Instrument retrieval and response evaluation against curated golden datasets; ground answers in certified semantic layers and knowledge graphs where appropriate, and design graceful abstention behavior.
Define prompt and instruction management standards, including versioning, templating, context budgeting, and regression testing.
Own the AI gateway, multi-model routing, model benchmarking and selection discipline, and model re-testing as the landscape changes.
Establish an evaluation harness as a launch gate, including offline and online evaluation, calibrated LLM-as-judge, A/B and shadow testing, and minimum quality thresholds.
Set evidence-based guidance on prompt engineering versus fine-tuning or customization, and define caching strategy.
Architect agent orchestration across agents, tools, retrieval, and human review; define domain-agent scope and reusable skills and tools.
Design memory and context-management policies and enterprise standards for tool/function calling and MCP server contracts.
Establish agent guardrails for authorization, permissions, approvals, escalation, sandboxing, fallback, safe failure, tracing, regression testing, audit, and incident response.
Define the agent registry, tool catalog, memory and state services, and the lifecycle from prototype to monitored production asset; apply agentic patterns to engineering practices.
Define responsible AI standards for fairness, transparency, explainability, and human oversight in alignment with enterprise data governance.
Establish AI security patterns for prompt injection, data exfiltration, jailbreaks, and model or supply-chain risk; partner with Data Governance on classification, stewardship, certified sources, and access controls.
Classify use cases by risk, drive attestation with Security, Legal, Privacy, and Risk, and lead technical reviews of third-party models and vendors.
Instrument tracing, telemetry, and drift signals; build cost attribution and make cost per successful answer a first-class metric.
Partner with platform and finance teams to connect AI investment to measurable outcomes and unit economics.
Define the Azure-native platform stack and lifecycle with MLOps and Engineering, including Azure AI Foundry, Azure ML, AKS, registries, CI/CD, landing zones, identity and network patterns, capacity, and LLMOps/AgentOps.
Partner with Engineering and Architecture on data readiness, certified metrics, lineage, data contracts, and converting reference patterns into shipped services and sanctioned builder environments.
Raise engineering and solution-architecture standards through design reviews, system-level reasoning, and mentorship without direct reporting relationships.
Communicate architecture trade-offs to senior technology leaders and peer architects, and lead cross-team decisions with Microsoft, Databricks, AI vendors, and systems integrators.
Write architecture decision records, reference patterns, and reusable accelerators, and develop proof-of-concept reference implementations to help teams move solutions into production.