AI Incubation Forward Deployment Engineer

DUTIES & RESPONSIBILITIES AI Technology Research & Exploration   • AI Technology Radar: Continuously track the latest developments in AI, including LLMs, multimodal models, AI agents and multi-agent frameworks, RAG, Model Context Protocol (MCP), and AI coding tools; maintain an AI technology radar for AZ China and share regular insights and trend briefings.  • Framework & Product Evaluation: Research, benchmark, and evaluate AI frameworks, platforms, and products (e.g., LangChain/LangGraph, LlamaIndex, Dify, Semantic Kernel, Microsoft Copilot Studio, Azure OpenAI, and leading China LLMs such as Qwen and DeepSeek), assessing capability, maturity, cost, security, compliance, and availability in China.  • Hands-on Experimentation: Set up AI sandboxes and run rapid technical experiments to validate new technologies against enterprise scenarios, documenting findings, limitations, and lessons learned.  • Technology Recommendations: Provide recommendations on AI technology selection and adoption to enterprise architecture and IT leadership, aligned with AZ Global AI strategy, platforms, and standards.  Business-Driven AI Opportunity Discovery   • Forward-Deployed Business Engagement: Work side by side with business functions (e.g., Commercial, Medical, R&D, Supply Chain, Finance, HR, Compliance) to understand their workflows, pain points, and data, and uncover where AI can create meaningful impact.  • Use Case Identification & Prioritization: Identify and shape high-value AI use cases; assess business value, technical feasibility, data readiness, and risk; and maintain a prioritized AI innovation pipeline.  • Alignment with Enterprise AI Direction: Align the incubation portfolio with the AZ China and Global enterprise AI roadmap, reusing global AI platforms and assets as the default and localizing for China only where necessary.  • Value Hypothesis & Success Criteria: Define clear value hypotheses, KPIs (e.g., productivity gains, cycle-time reduction, quality improvement, user adoption), and go/no-go criteria for each innovation before incubation starts.  AI Innovation Incubation (0 to 1)   • Rapid Prototyping & PoC: Design and build PoCs and MVPs hands-on within weeks, such as AI agents, RAG-based knowledge assistants, intelligent document processing, and AI-enabled business workflows.  • Solution Design & Architecture: Design end-to-end AI solution architectures, covering model selection, prompt and context engineering, retrieval, tool and agent orchestration, and integration with enterprise systems and data (e.g., SSO, APIs, M365, Workmate).  • Pilot & Iteration: Deploy pilots with real users, collect feedback and usage data, and iterate quickly to refine user experience, accuracy, and business fit.  • Evaluation & Measurement: Establish evaluation frameworks and datasets to measure answer quality, accuracy, hallucination, latency, and cost, and quantify the business value delivered by each pilot.  • Go/No-Go Decisions & Reporting: Present incubation outcomes to business and IT stakeholders and recommend whether to scale, pivot, or stop each innovation based on evidence.  AI Innovation Scale-up & Delivery Enablement (1 to 10)   • Transition to Development Teams: Hand over validated innovations to development teams with complete solution architecture, code, documentation, evaluation sets, and known risks, ensuring a smooth transition from prototype to product.  • Co-Development & Technical Guidance: Support and guide development teams through engineering hardening and productionization, including performance, scalability, security, observability, and LLMOps practices, and step in hands-on to solve critical technical challenges.  • Reusable Assets & Reference Patterns: Build and maintain reusable AI components, templa