Cloud AI Application Administrator

Must have: Agentic AI Architecture & Workflow Design, Product-Level Architecture Ownership (C4 + ADRs + NFRs), AI Evaluation Safety & Auditability. Azure Cloud Architecture & Hands-On Prototyping. C#/.NET, Python/TypeScript for POCs, RAG / Retrieval Architecture & Knowledge-Driven Foundations and Kafka/Confluent About the Role This is a hands-on product architecture role at the frontier of AI-driven audit transformation. You will partner directly with engagement teams to modernize the audit workflow using agentic, knowledge-driven capabilities - not bolt-on AI tools, but deeply embedded intelligent systems that enhance auditor judgment while preserving full control, explainability, and review-ready provenance. You will own product-level architecture across key audit workflow surfaces - planning, risk assessment, procedures, and completion - along with the knowledge and context layers that power them and the guardrails that make agentic execution trustworthy in a regulated environment. This is not a strategy-only or slides-only role. You will prototype, contribute artifacts to repositories, author architecture decision records, and be directly accountable for shipping production-ready agentic workflows that auditors can rely on. What You Will Do

Product Architecture Ownership Own end-to-end product-level architecture for audit engagement workflows and agentic experiences, producing context/container diagrams, runtime and sequence flows, data models, and integration contracts Define how AI agents operate as "experienced audit assistants" - augmenting auditor judgment while keeping auditors explicitly in control through approval gates, audit trails, and enforceable policy constraints Ensure interoperability patterns for future ecosystem integrations using standard protocols, while maintaining privacy, security, and data residency requirements Knowledge & Context Layer Design Design knowledge-driven foundations covering document and data ingestion, indexing pipelines, knowledge graph and vector store architecture, and evidence binding to support grounded, reviewable agent outputs Specify agent orchestration patterns - Planner Executor and Manager?Executor models - aligned to specific workflow steps, with clearly defined tool access boundaries and least-privilege execution controls Evaluation, Quality & Observability Establish structured evaluation loops: offline test sets, regression suites, drift monitoring, and red-team scenarios tailored to audit defensibility requirements Define quality metrics relevant to audit-grade outputs and set performance targets (p95 latency, cost budgets) with live observability dashboards Build rapid spikes and POCs to de-risk agent workflows, retrieval and evidence patterns, and performance assumptions; document all architectural decisions via ADRs Cross-Functional Delivery Partnership Partner closely with Product Owners, UX, and Engineering Leads to translate firm and client feedback into buildable epics and stories with measurable acceptance criteria and clearly defined Non-Functional Requirements (NFRs) Serve as the architectural connective tissue between product vision and engineering execution - ensuring what gets built is both technically sound and workflow-native

• 8+ years in software engineering with 3+ years in product or application architecture for SaaS or distributed systems
• Hands-on Azure experience - AKS/Kubernetes, Key Vault, Blob/ADLS, networking/ingress, observability - with practical application of Well-Architected Framework principles
• Strong documentation and governance discipline: C4/Mermaid diagrams , ADR authoring, and NFR definition with measurable acceptance criteria
• Proven hands-on prototyping ability - comfortable building spikes and POCs, and reverse-engineering modern .NET services where needed
• Demonstrated experience implementing AI features with explicit evaluation, safety, and auditability - including logging, traceability, and human-in-the-loop (HITL) approval gates, * Experience designing agentic systems in regulated domains - audit, finance, legal, or healthcare - with defensible reasoning trails and compliance-grade audit logs
• RAG / retrieval expertise - chunking strategies, citation-first prompt design, index schema design, and evidence binding to immutable stores
• Knowledge graph modelling for workflow traceability - mapping entities, risks, procedures, evidence, and findings for downstream navigation and review
• Eventing and streaming experience (Kafka/Confluent) and workflow orchestration patterns - idempotency, retries, durable commands
• Familiarity with Microsoft Foundry Agent Service / MAF and Azure AI Search Core Competencies Systems thinking - ability to reason across the full stack and make pragmatic architectural trade-offs under real constraints Bias for action with design rigor - moves fast but leaves high-quality traces: docs, code, and ADRs Communication range - able to shift fluently between executive narrative and code-level technical guidance Safety-minded AI design - designs with explicit agent boundaries, evaluators, and failure handling baked in from the start, not added as an afterthought, Required: Bachelor's degree in Computer Science or a related field This role is open to W2 or those seeking Corp-Corp employment. The salary range for this role is $106 - $114k. For corp-Corp rates please contact the recruiter. In addition to other benefits, OpenKyber offers a comprehensive benefits package, with OpenKyber covering 65% of the medical, dental, and Vision Premiums for employees, their spouses, and dependent children enrolling in the OpenKyber-provided plans.