Senior AI Engineer

Overview

Key Responsibilities

• Use case technical ownership: Own the technical delivery of one or more AWA use cases from discovery through production; define the Agent Profile, Use Case Manifest, Data Manifest, and Tool Authorisation Matrix for each use case.
• Architecture and design: Design LangGraph agent graph topologies for complex multi-agent pipelines including Supervisor-Worker fan-out, parallel execution, state compression, and cross-document reconciliation (3-way match).
• Confidence calibration and HITL thresholds: Lead the AI Metric Baselines calibration process: measure ECE, derive HITL routing thresholds empirically, present threshold recommendations to business and risk teams, and own the signed AI Metric Baselines table.
• Test strategy ownership: Own and execute the full AWA IDP Test Strategy for your use cases: all four testing bands (Model/Prompt/Agent/Security, Adversarial, Integration, UAT) through to sign-off.
• Agent Regression Testing: Maintain the Golden Dataset and Regression Dataset for your use cases; configure CI/CD regression pipeline gates; review regression results after every model or prompt change.
• Mentoring and code review: Mentor AI Engineers; conduct substantive technical code reviews of agent graphs, prompts, and MCP tool implementations; maintain AWA coding standards.
• Stakeholder engagement: Present model quality metrics, STP rates, and HITL rates to Ops leads and business owners; recommend threshold adjustments with data-backed rationale.
• New capability adoption: Evaluate and prototype new AI capabilities (new Azure AI Foundry models, new LangGraph features, new RAG patterns) for incorporation into the AWA accelerator catalogue.

Required Skills and Experience

• Technical — Essential
• 5+ years Python development; 2+ years in production AI/ML engineering
• Deep LangGraph or equivalent agentic framework experience: graph topology, state management, conditional edges, checkpoint-based recovery
• LLM evaluation expertise: F1 measurement against ground truth, Expected Calibration Error (ECE), hallucination detection, confidence calibration
• Prompt engineering at scale: versioned prompts, sensitivity analysis, few-shot dataset curation, PMS governance
• RAG architecture: hybrid retrieval (BM25 + vector), re-ranking, chunk citation, grounding verification
• Azure AI Foundry or equivalent LLM inference platform: PTU, PAYG, token budget management, APIM integration
• Workflow orchestration: Orkes Conductor or Temporal — HITL wait tasks, retry policies, compensating transactions
• CI/CD pipeline integration for ML: Azure DevOps, MLflow experiment tracking, model registry promotion
• Technical — Advantageous
• Computer vision model deployment and evaluation (object detection — TPR/TNR/IoU)
• Azure AI Document Intelligence: form recogniser, custom models, layout analysis
• Kubernetes / AKS: workload identity, namespace isolation, KEDA autoscaling
• Apache Iceberg, DuckDB, and data contract patterns
• Banking domain knowledge: cheque clearing, trade finance, KYC, AML

Qualifications

• Bachelor's degree in Computer Science, AI/ML, or equivalent; Master's preferred
• Azure AI Engineer Associate or equivalent certification
• Demonstrable production AI deployments — case studies or public work highly valued
• Pay: AED21,000.00 - AED30,000.00 per month