AI Solution Architect

NCS is a leading technology services firm that operates across the Asia Pacific region in over 20 cities, providing consulting, digital services, technology solutions, and more. We believe in harnessing the power of technology to achieve extraordinary things, creating lasting value and impact for our communities, partners, and people. Our diverse workforce of 15,000 has delivered large-scale, mission-critical, and multi-platform projects for governments and enterprises in Singapore and the APAC region.   As an AI Solution Architect, you will be the technical authority responsible for designing, architecting, and guiding the delivery of Agentic AI systems within the Autonomous Networks Programme. You will work closely with the Project Director, Network SMEs, OSS engineers, NOC operations teams, and the Central AI Infrastructure Solution Architect to translate network operations challenges into production-grade AI agent solutions that are safe, explainable, trusted, and operationally effective. This is a hands-on role requiring both strong architecture leadership and practical engineering capability, including prototyping, coding, system integration, and technical review. What will you do? Agent Architecture & Design • Lead the end-to-end architecture of Agentic AI systems, covering data ingestion, reasoning, decision-making, action execution, and audit logging. • Define agent scope, decision boundaries, tool specifications, and autonomy levels in alignment with programme stakeholders. • Design multi-agent orchestration capabilities such as event routing, resource locking, inter-agent handoff contracts, prompt caching, and shared state management. • Architect RAG pipelines for knowledge and diagnosis agents, including corpus design, chunking strategy, embedding model selection, retrieval design, and evaluation methodology. • Design agentic workflows, evaluation patterns, hybrid search, reranking strategies, benchmarking approaches, and golden datasets. • Develop prompt templates, output schemas, hallucination guardrails, and confidence scoring mechanisms for LLM-based agents. • Produce and own technical design artefacts including architecture decision records (ADRs), sequence diagrams, data flow diagrams, and agent logic flowcharts. OSS Integration & Data Engineering • Adopt and apply Model Context Protocol (MCP) and Agent-to-Agent protocol patterns as governed extension mechanisms for enterprise AI agents. • Design and build data pipeline layers that normalise multi-vendor, multi-format network telemetry into usable schemas for AI reasoning. • Work with the Network Solution Architect to define integration patterns between AI agents and OSS platforms such as NMS, EMS, ServiceNow ITSM, inventory, and provisioning systems through REST APIs. • Design event streaming architecture, such as Kafka or equivalent, to support near real-time alarm and telemetry processing aligned to NOC latency requirements. • Collaborate with Data Engineering teams to improve the quality of alarm, ticket, and inventory data that supports agent accuracy and reliability. Safety, Guardrails & Risk Architecture • Design and implement guardrails for all agents, including hard limits, human confirmation flows, blast radius controls, and escalation logic. • Lead blast radius assessments with Network SMEs and Risk teams, documenting worst-case impacts and defining mitigating controls. • Design and validate rollback procedures for agent-initiated actions, including KPI-based automated rollback triggers. • Ensure solution architectures comply with IMDA requirements, AIVerify expectations, InfoSec policies, and relevant change governance processes. Observability & Evaluation • Design decision audit trail schemas to ensure all agent actions and decisions are traceable, explainable, and reviewable. • Define continuous evaluation approaches for production agent traces using LLM-as-judge and human review samples. • Build or guide the development of Agent Operations Dashboards for NOC teams, showing agent health, decision activity, accuracy, and exceptions. • Define agent evaluation frameworks across factual accuracy, reasoning quality, retrieval performance, planning quality, tool-call accuracy, and confidence thresholds. • Design drift detection mechanisms to identify output quality degradation before operational impact occurs. Delivery & Stakeholder Collaboration • Partner with the Project Director on design planning, sprint planning, design governance, and build readiness. • Work with Network SME owners to validate architectural decisions against real operational workflows and constraints. • Lead technical walkthroughs with NOC management and Central AI Infrastructure leadership to build confidence in the proposed architecture. • Provide mentorship and technical guidance to AI/ML engineers and integration engineers through code reviews, architecture reviews, and design support. • Translate technical decisions into clear, concise, non-technical summaries for programme leadership and steering committees. The ideal candidate should possess: Required / Mandatory experience • Minimum 7 years of experience in AI/ML engineering or application solution architecture. • At least 1 year of hands-on experience in production Agentic AI systems, beyond proof-of-concept or prototype work. • Demonstrated experience designing and deploying AI agents that use external tools, manage state, and operate in production environments. • Hands-on experience with LLM APIs such as OpenAI or Anthropic Claude, or with open-source LLMs on vLLM such as Qwen, Mistral, or DeepSeek. • Strong experience in prompt engineering, evaluation design, and cost management for production AI systems. • Proven experience designing and implementing RAG pipelines with vector databases such as Milvus, Pinecone, Weaviate, Chroma, pgvector, or equivalent. • Experience integrating AI systems with enterprise operational platforms such as ITSM, monitoring, streaming, or telemetry systems. • Ability to write code, build prototypes, and review pull requests as part of hands-on delivery. Highly desirable experience • Experience in Telco, utilities, or other 24×7 critical infrastructure environments. • Familiarity with OSS platforms and protocols such as NMS, EMS, network alarm management, NETCONF/YANG, or gNMI streaming telemetry. • Experience designing AI systems under regulatory, audit, security, and formal risk governance requirements. • Experience with network automation platforms such as Cisco NSO, Nokia CloudBand, or ONAP. • Familiarity with TM Forum Autonomous Network Maturity Model (L0–L5). • Working knowledge of frameworks such as LangChain and LangGraph. Desirable certifications • Microsoft Certified Agentic AI Business Solutions Architect • AWS Certified Generative AI Developer – Professional • AWS Machine Learning Specialty • Google Professional Machine Learning Engineer Education qualification • Degree in Computer Science, Electrical Engineering, AI/ML, or a related technical field. • A postgraduate qualification is advantageous, but not required if supported by strong demonstrated experience.