JD4 — Senior AI Engineer / Agentic AI Engineer (Banking, Microsoft Azure)

Source. This is the job description the candidate received via an interview follow-up, cleaned, de-duplicated, and validated. The original arrived as run-together HR text (words fused: "applicationBuild", "Agentic A", "English / Arabic" with a missing paren, the three "MUST/THE MUST" anchors repeated out of order). Nothing material was changed — only segmentation, spelling, and ordering were repaired so each requirement is legible and individually checkable. A claim-by-claim validation table follows in §5.

This is a fourth distinct role in the AI Specialist/ book, alongside JD1 — Air-Gapped Digital Twin, JD2 — Inference Optimization, and JD3 — AppliedAI / Opus. It is the only Azure-native, agentic, regulated-banking role of the four, so it gets its own dedicated curriculum: ➡ jd4-azure-agentic-banking-track/.

Table of Contents

• 1. The Role (cleaned)
• 2. Key Responsibilities (cleaned)
• 3. Mandatory Skills (cleaned)
• 4. Good to Have (cleaned)
• 5. Validation: is every requirement real and current?
• 6. What this JD really is (the subtext)
• 7. Where each requirement is covered in this track

1. The Role (cleaned)

We are hiring a highly skilled Senior AI Engineer / Agentic AI Engineer with strong experience delivering enterprise-scale AI solutions in Banking or Financial Services. The ideal candidate has hands-on expertise building Agentic AI systems, Generative AI pipelines, and production-grade Conversational AI / RAG platforms on Microsoft Azure.

You will work on real-world banking use cases such as Conversational AI, Document Intelligence, Agentic Workflow Automation, and Multimodal AI systems serving large user bases.

The three "THE MUST" anchors (the recruiter capitalised these — they are the screening gates):

1. Design and implement Agentic AI architectures for banking-grade enterprise applications.
2. Build scalable Generative AI / RAG pipelines using Azure OpenAI and Azure AI Search.
3. Strong, demonstrable Azure experience across: Azure OpenAI, Azure AI Search, Azure AI Document Intelligence, Azure AI Vision, and Azure AI Foundry. Azure OpenAI Service specifically is called out as the single non-negotiable ("THE MUST").

2. Key Responsibilities (cleaned)

1. Design and implement Agentic AI architectures for banking-grade enterprise applications.
2. Build scalable Generative AI / RAG pipelines using Azure OpenAI and Azure AI Search.
3. Deliver production-grade AI systems with high availability, security, and performance.
4. Implement multimodal and multilingual AI solutions (Text + Vision, including English / Arabic).
5. Develop document-processing pipelines using Azure AI Document Intelligence.
6. Build Computer Vision workflows using OpenCV.
7. Develop REST APIs using FastAPI and deploy using Docker and Azure Containers.
8. Design embedding, vector search, and hybrid retrieval strategies.
9. Integrate Azure Cosmos DB (Document/NoSQL).
10. Implement CI/CD pipelines and Git-based workflows.
11. Collaborate closely with business stakeholders and engineering teams to deliver end-to-end AI solutions.

3. Mandatory Skills (cleaned)

• 3+ years hands-on in AI / ML / Generative AI / Agentic AI.
• Prior experience implementing AI in Banking or Financial Services, delivering enterprise-scale, production-grade systems.
• Strong Python and SQL.
• Microsoft Azure, specifically: Azure OpenAI, Azure AI Search, Azure AI Document Intelligence, Azure AI Vision, Azure AI Foundry.
• Agentic / LLM frameworks: LangGraph, LangChain.
• Advanced Prompt Engineering.
• Computer Vision with OpenCV.
• API development with FastAPI.
• Git, Docker, Azure Container Services.
• Azure Cosmos DB (Document/NoSQL).
• CI/CD pipelines.

4. Good to Have (cleaned)

• Open-source AI: YOLO models, Deep Learning, NLP.
• Agent orchestration frameworks: Semantic Kernel, Microsoft Agent Framework.
• LLM evaluation metrics and frameworks (RAG evaluation, custom benchmarks, etc.).
• MCP (Model Context Protocol).
• Orkes (agentic-workflow / durable-orchestration tool, built on Netflix Conductor).
• Streamlit or Flask for rapid PoC development.
• Microsoft Certification: Azure AI Engineer Associate (AI-102).

5. Validation: is every requirement real and current?

Every named technology was checked against its current (2026) product reality. The point: nothing in this JD is fictional or deprecated, but several items were renamed by Microsoft, and you must use the current names in interview to signal you actually use the platform.

Verdict: the JD is coherent, current, and internally consistent. It describes a real, mainstream 2026 Azure-native agentic-AI engineering role in regulated banking. No red flags. The only stale items are naming (Cognitive Search → AI Search, Form Recognizer → Document Intelligence) and the imminent AI-102 retirement — both of which you should know and mention, because doing so proves you live on the platform.

6. What this JD really is (the subtext)

Decoded, this role wants one person who can do all five of the following, in a regulated, audited, multilingual (Arabic/English) banking context:

1. Agentic AI architect — design multi-step, tool-using, stateful agents (LangGraph / Agent Framework / Foundry Agent Service) that do work, not just chat.
2. RAG/GenAI platform builder — Azure OpenAI + Azure AI Search hybrid retrieval at scale, grounded and evaluated.
3. Document Intelligence engineer — turn statements, KYC docs, cheques, trade-finance paperwork into structured data.
4. Multimodal/vision engineer — OpenCV/YOLO + Azure AI Vision, Arabic OCR, signature/ID/cheque processing.
5. Production platform engineer — FastAPI + Docker + Cosmos DB + CI/CD + security/availability, on Azure.

The connective tissue the JD only implies but every banking interviewer will probe: security, compliance, data residency, auditability, groundedness/hallucination control, PII handling, and cost — because this is a bank. That implied layer is where senior candidates win or lose, and it gets its own knowledge module (Banking Domain, Security & Responsible AI).

7. Where each requirement is covered in this track

➡ Start at the track README. If the interview is imminent, jump straight to the Interview-Day Battle Plan.