AI-Native Engineer (FDE)/12 months contract/Insurance

Job Title: Forward Deployment Engineer / AI-Native Engineer

Department: Enterprise AI Engineering & Digital Transformation

Location: Hong Kong

Position Overview

We are seeking an AI-Native Forward Deployment Engineer (FDE) to design and deploy enterprise-grade AI systems within the Hong Kong insurance sector. Unlike traditional software roles focused on CRUD APIs and standard microservices, this role centers on AI System Design: building agentic reasoning workflows, optimizing context windows and token budgets, enforcing strict regulatory guardrails, and managing non-deterministic AI performance at scale.

You will operate at the intersection of production AI engineering and client delivery, translating complex insurance workflows (Claims Adjudication, Underwriting Automation, Policy Processing) into resilient, scalable AI infrastructure integrated directly into core enterprise backends.

Key Responsibilities

• AI System Design & Optimization: Architect non-deterministic, scalable AI systems. Implement model-routing architectures (balancing local SLMs with frontier LLMs), semantic caching, and speculative decoding to optimize token expenditure, execution latency, and cost.

• Agentic RAG & Multi-Agent Workflows: Design and deploy multi-agent reasoning workflows (e.g., using LangGraph, AutoGen) and layout-aware hybrid RAG pipelines (dense/sparse vector search, dynamic re-ranking, and GraphRAG) to automate complex medical claims, underwriting, and policy documents.

• AI Guardrails & Safety Engineering: Build robust input/output guardrail layers (utilizing NeMo Guardrails, Guardrails AI, or custom validators) for real-time PII masking, hallucination prevention, and jailbreak defense, ensuring 100% compliance with Hong Kong’s Personal Data (Privacy) Ordinance (PDPO) and Insurance Authority (IA) standards.

• Enterprise Integration: Modernize legacy enterprise touchpoints by connecting non-deterministic AI pipelines with core policy administration platforms (e.g., AS400, Life/400) and enterprise data warehouses.

• LLMOps & Non-Deterministic Evaluation: Establish production evaluation harnesses using LLM-as-a-Judge methodologies to continuously benchmark retrieval accuracy, semantic drift, and hallucination rates.

• Field Deployment: Partner directly with front-line business stakeholders (Underwriters, Claims Adjusters, Operations) to rapidly prototype, stress-test, and deploy high-impact AI solutions in short iterative cycles.

Key Qualifications & Technical Expertise

• AI System Architecture: 3+ years of experience engineering production AI systems, with a deep understanding of non-deterministic execution, context window management, and token optimization.

• Core Technical Stack: Advanced proficiency in Python, SQL, and multi-agent orchestration frameworks ( LangGraph, LangChain, LlamaIndex).

• Vector Databases & Search: Hands-on experience architecting vector search solutions using Pinecone, Qdrant, Milvus, or pgvector, including hybrid search mechanics and metadata filtering.

• AI Safety & Compliance: Proven track record of configuring guardrail systems, PII redacting pipelines, and security controls for sensitive insurance/financial data.

• Infrastructure & Cloud: Experience with containerization ( Docker, Kubernetes) and enterprise cloud AI platforms ( Azure AI, AWS Bedrock, or GCP).

Argyll Scott Asia is acting as an Employment Business in relation to this vacancy.