New opportunity

AI Solution Architect, Vice President – Corporate Functions Technology

State Street · Boston, United States of America

About this role

State Street seeks a hands-on AI Solution Architect, Vice President, to lead AI strategy and architecture for Internal Audit. The role designs, builds and operates enterprise-scale systems delivering conversational, predictive and generative capabilities to improve audit quality, productivity and insights.

Responsibilities

Set architecture standards and lead AI platform design across LLMs, machine learning models, enterprise data platforms and core systems. Develop reusable patterns for conversational assistants, generative insights, predictive analytics and continuous auditing.
Design multi-step, agent-based workflows using AI orchestration frameworks such as LangChain or equivalent cloud-native services. Incorporate RAG over structured and unstructured data, guardrails, validation and hallucination mitigation. Own model selection, evaluation and benchmarking, balancing performance, cost, latency, explainability and risk.
Deliver production AI systems, including deployments, APIs, orchestration pipelines and enterprise integrations. Establish LLMOps/MLOps practices for versioning, monitoring, evaluation, logging, rollback and cost control; maintain availability, scalability and resilience.
Embed model risk management, privacy, security, bias detection, explainability and auditability. Work with Internal Audit, Risk, Compliance, Legal, data engineering, security and infrastructure teams. Turn audit challenges into use cases such as control-testing automation, issue identification, narrative generation and continuous auditing; guide proofs of concept toward production.

Qualifications and preferences:

A bachelor’s or master’s degree in Computer Science, Engineering, Data Science or a related field is listed under education and preferred qualifications, alongside 10+ years designing and delivering enterprise software or data platforms, including 3+ years focused on AI/ML solutions. The posting calls for hands-on LLM, prompt engineering, embeddings, RAG and agent experience; Azure or AWS and microservices experience; MLOps/LLMOps, data engineering, API and distributed-systems knowledge; and security, privacy and regulatory design capability. Communication, mentoring and cross-functional partnership are valued.
Nice to have: experience with Internal Audit, Legal or Risk functions; AI governance or model risk frameworks; vector databases, semantic search, data catalogs, process mining, control-testing automation or continuous auditing; and cloud or AI/ML certifications.

The posting lists Boston and Clifton, New Jersey. The annual salary range is $120,000–$217,500 for the primary location and may differ elsewhere. Benefits include a matched 401(k), insurance, paid time off, an employee assistance program and eligibility for performance-based awards.

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

Skills for this role

AI architectureGenerative AILarge language modelsMachine learningPrompt engineeringEmbeddingsRetrieval-augmented generationAI agentsLangChainAI orchestrationModel evaluationBenchmarkingGuardrailsHallucination mitigationLLMOpsMLOpsModel deploymentAPIsAzureAWSMicroservicesData engineeringDistributed systemsAI governanceModel risk managementData privacySecurityBias detectionExplainabilityStakeholder collaboration Web Communication?

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

Set the Internal Audit AI architecture
The role owns end-to-end architecture and standards for platforms integrating LLMs, machine learning, enterprise data and core systems. Reusable patterns must support conversational, predictive, generative and continuous-auditing use cases.
Design and evaluate controlled AI workflows
The posting calls for multi-step, agent-based workflows with retrieval over structured and unstructured data, guardrails and validation. It also assigns ownership of model evaluation and benchmarking across performance, cost, latency, explainability and risk.
Take AI systems into reliable production
The architect leads deployments, APIs, orchestration pipelines and enterprise integrations, while establishing LLMOps/MLOps practices for monitoring, logging, versioning, rollback and cost control.
Build governance into the design
The role embeds model risk management, privacy, security, bias detection, explainability and auditability, working with Internal Audit, Risk, Compliance and Legal on policy and regulatory alignment.
Turn audit needs into viable use cases
The posting names control-testing automation, issue identification, narrative generation and continuous auditing. The architect guides proofs of concept toward production and communicates architectural decisions across business and technology teams.

Likely challenges to think through

Balancing model quality against operating constraints
Preparation hypothesis: The best-performing model may not meet cost, latency, explainability or risk needs. Prepare a decision framework with benchmarks, acceptance criteria and examples of when a simpler approach is preferable.
Making retrieved and generated audit outputs dependable
Preparation hypothesis: Combining structured and unstructured sources in agent workflows may make unsupported outputs harder to detect. Prepare to explain retrieval checks, validation, guardrails and how outputs can be traced for review.
Moving from proof of concept to governed operations
Preparation hypothesis: A promising audit use case may be difficult to scale while preserving privacy, resilience and auditability. Prepare a staged production plan covering integrations, monitoring, rollback and cross-functional review.

Practice your answers

Technical practice

How would you architect a reusable AI platform for Internal Audit across conversational, predictive and generative use cases?

What to coverThe posting calls for enterprise-scale patterns integrating LLMs, machine learning models, data platforms and core systems. Suggested preparation: Explain component boundaries, data flows and architectural trade-offs that allow reuse without weakening security or reliability.

How would you design RAG over structured and unstructured audit data?

What to coverThe posting specifically calls for RAG across both data types and lists semantic search and data catalogs as nice-to-have knowledge. Suggested preparation: Discuss data preparation, retrieval and embedding choices, access controls, source attribution and how you would test retrieval quality.

When would you use a multi-step agent workflow rather than a simpler orchestration pipeline?

What to coverThe role includes agent-based workflows using LangChain or equivalent cloud-native capabilities. Suggested preparation: Compare flexibility with complexity, latency and risk, and explain where validation and human review would fit.

How would you reduce hallucinations in an audit-facing generative AI workflow?

What to coverThe posting explicitly calls for guardrails, validation layers and hallucination mitigation. Suggested preparation: Cover grounded prompts, checks against retrieved evidence, handling unsupported answers and ways to measure whether the safeguards work.

How would you select and benchmark models for an Internal Audit use case?

What to coverThe role owns model selection and evaluation while balancing performance, cost, latency, explainability and risk. Suggested preparation: Define a representative test set and acceptance criteria, then explain how the trade-offs would affect your choice.

How would you take an AI proof of concept into a production enterprise integration?

What to coverThe posting calls for model deployments, APIs, orchestration pipelines, enterprise integrations and a clear path from experimentation to production. Suggested preparation: Describe interface design, data dependencies and cloud or microservices decisions, including what you would validate before release.

What LLMOps and MLOps practices would you establish for this platform?

What to coverThe posting names versioning, monitoring, evaluation, logging, rollback and cost controls. Suggested preparation: Explain how these practices work together to detect quality or cost changes, trace a release and restore a prior version.

How would you troubleshoot a production AI service that has become slow or unreliable?

What to coverThe role is responsible for availability, scalability, performance, resilience and reliability. Suggested preparation: Walk through how you would isolate model, retrieval, orchestration and integration bottlenecks, then weigh mitigations against answer quality and cost.

How would you embed governance and model risk controls in the architecture?

What to coverThe posting names privacy, security, bias detection, explainability and auditability, alongside model risk management. Suggested preparation: Explain where safeguards and evidence capture belong in the system and how you would assess their effectiveness with relevant control partners.

How would you evaluate control-testing automation as an AI use case?

What to coverControl testing automation is one of the posting's example Internal Audit use cases. Suggested preparation: Define the task and evidence needed, compare automation with human review, and propose measures for correctness, auditability and production readiness.

Behavioural practice

When have you influenced an architectural decision across multiple initiatives?

What to coverThe role sets AI architecture standards and guides decisions across Internal Audit Technology. Suggested preparation: Use a real situation, your actions to resolve competing priorities and the result; distinguish your contribution from the team's.

When have you worked through a disagreement with risk, security or business stakeholders?

What to coverThe posting emphasizes partnership with Internal Audit, Risk, Compliance, Legal and technology teams. Suggested preparation: Describe a real situation, how you listened and negotiated a technical approach, and the resulting decision or outcome.

When have you turned an ambiguous business problem into a viable technical use case?

What to coverThe role translates audit and risk challenges into AI use cases and guides proofs of concept toward production. Suggested preparation: Give a real situation, the actions you took to define value and feasibility, and the result, including any reason you chose not to proceed.

When have you explained a complex technical trade-off to a non-technical audience?

What to coverClear communication with technical and non-technical audiences is valued in the posting. Suggested preparation: Recount a real situation, how you made the options and risks understandable, and what decision or outcome followed.

When have you mentored others or guided adoption of an emerging technology?

What to coverThe posting values mentoring and pragmatic adoption of evolving AI technologies. Suggested preparation: Describe a real situation, the guidance or evaluation you provided, and the result without overstating your individual impact.

Sharpen your knowledge

Topics to prepare

Enterprise AI architecture
The posting calls for reusable, production-grade architecture integrating LLMs, ML models, enterprise data platforms and core systems. Revise by sketching a reference architecture for an Internal Audit assistant and explaining its scalability, resilience and integration trade-offs.
RAG and agent workflow design
The role calls for multi-step agent workflows, RAG across structured and unstructured data, validation and hallucination mitigation. Revise by tracing an audit question from retrieval through generation to validation, identifying where a workflow should stop or request review.
Model selection and benchmarking
The architect owns evaluation frameworks balancing performance, cost, latency, explainability and risk. Revise by defining a comparison matrix for two hypothetical models and explaining which trade-offs would change the choice.
AI governance and controls
The posting requires model risk management and safeguards for privacy, security, bias, explainability and auditability. Revise by mapping each concern to a design control and the evidence an auditor could inspect.
Internal Audit use cases
The role translates audit challenges into use cases, including control testing automation, issue identification, narrative generation and continuous auditing. Revise by choosing one listed use case and defining its inputs, intended output, limitations and measures of audit value.
Production operations principles
The posting specifies LLMOps/MLOps, availability, reliability, monitoring, rollback and cost control. Revise by describing how a model or prompt change would be versioned, evaluated, released, monitored and rolled back.

Skills to revise

Build a grounded workflow
The posting asks for hands-on prompts, embeddings, RAG and agent-based orchestration using LangChain or equivalent capabilities. As a practice exercise, assemble a small workflow over sample structured and unstructured audit data that retrieves evidence, drafts an answer and validates its grounding.
Benchmark model behavior
The role includes model evaluation and benchmarking. Practice creating test cases for a listed audit use case, recording answer quality, unsupported claims, latency and cost, then explaining a model-selection decision.
Design an enterprise integration
The posting calls for deployments, APIs, enterprise integrations, Azure or AWS, and microservices architecture. Practice drawing an API and deployment design connecting an AI service to an enterprise data platform; annotate access boundaries and failure handling.
Run an LLMOps change drill
Versioning, monitoring, logging, evaluation, rollback and cost controls are explicit responsibilities. Practice writing a release checklist for a changed prompt or model, including what to log, which signals trigger rollback and how to verify recovery.
Present an audit use-case proposal
The role partners with Internal Audit, Risk, Compliance, Legal and technology teams and guides proofs of concept toward production. Practice a brief proposal for one listed use case, stating expected audit benefit, required data, governance questions and production-readiness criteria.

Your preparation checklist

Map evidence to the role
Create a one-page inventory of relevant enterprise software or data-platform work and AI/ML work, with dates, your responsibilities, and verifiable outcomes. The posting lists 10+ years and 3+ years respectively; do not imply experience you cannot substantiate.
Prepare a production-system walkthrough
If you have a relevant example, outline the problem, architecture decisions, hands-on contributions, integrations, trade-offs, and measured results. Be ready to explain how you led technical decisions as well as delivery.
Practice an Internal Audit design exercise
Sketch a proposed architecture for one advertised use case, such as control-testing automation. Show data sources, APIs, cloud or microservice components, a path from proof of concept to production, and boundaries between AI output and audit judgment; label it as a proposed design.
Explain the RAG and agent workflow
Extend the design with retrieval over structured and unstructured data, multi-step orchestration, validation, guardrails, and hallucination mitigation. Prepare to justify where an agent is useful and where a simpler workflow would suffice.
Build a model-evaluation comparison
Prepare a sample scorecard comparing model options on task performance, cost, latency, explainability, and risk. State how you would benchmark outputs for the chosen audit use case rather than asserting that one model is best.
Outline a production operations plan
Draft a short runbook covering deployment, versioning, monitoring, logging, evaluation, rollback, cost controls, and availability. Identify the evidence you would use to show that the system remains reliable at enterprise scale.
Rehearse a governance and partnership scenario
Prepare a decision narrative showing how you would address privacy, security, bias, explainability, auditability, and model risk with Internal Audit, Risk, Compliance, Legal, data engineering, security, and infrastructure partners.
Write questions for the employer
Bring specific questions about priority Internal Audit use cases, available data and integrations, model-evaluation expectations, governance responsibilities, and what distinguishes a successful proof of concept from a production deployment.

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 of America. Matched to this role, not your profile.

Explore more jobs

Finding similar opportunities…