AI Architect
Description
AI Architect
Location : Seremban
Employment : 12Months Contract (Renewable)
Role Overview
We are seeking an experienced, hands-on AI Architect to lead the end-to-end architecture, design and implementation of an AI-driven lot disposition decision-support system within a semiconductor manufacturing environment.
The role will own the complete solution lifecycle, from High-Level Design (HLD) and Low-Level Design (LLD) through development, system integration, deployment and continuous improvement.
The solution will integrate manufacturing data from MES, SPC, yield management and test systems, combining traditional machine learning, LLMs and AI agents to support quality and product engineers in making faster, more accurate and evidence-based lot disposition decisions.
The AI Architect will design intelligent agents capable of classifying manufacturing holds, retrieving similar historical cases, assembling relevant evidence and recommending appropriate dispositions with confidence scores and traceable justifications.
This is a hands-on technical leadership position, requiring the ability to translate business and manufacturing requirements into production-ready architecture, guide AI engineers through implementation, and contribute directly to coding and technical reviews.
Given the automotive-grade quality requirements, the role carries significant responsibility for system reliability, explainability, auditability and safe failure. The successful candidate must ensure that AI recommendations remain controlled, transparent and subject to appropriate human approval.
You will work closely with the Integration Architect, Data Centre of Excellence (Data CoE), Manufacturing IT, Quality Engineering and Product Engineering teams across manufacturing sites.
Key Responsibilities
1. Solution Architecture and Design
- Collaborate with quality, product and process engineering teams to understand and document existing lot disposition workflows, including hold triggers, evidence requirements, decision criteria, Material Review Board (MRB) processes and approval hierarchies.
- Design the target AI-assisted disposition workflow, defining which activities can be automated, AI-assisted or retained as manual engineering decisions.
- Own the end-to-end solution architecture covering data ingestion, data processing, feature engineering, ML models, LLM-based agents, decision orchestration and audit capabilities.
- Develop and maintain comprehensive High-Level Design (HLD) documentation, including system context, component architecture, data flows, deployment architecture, integration landscape and non-functional requirements.
- Lead the development and sign-off of Low-Level Design (LLD), covering service interfaces, API contracts, data schemas, agent and tool definitions, model interfaces, prompt designs, sequence diagrams and exception-handling mechanisms.
- Evaluate and document key architectural decisions, including build-versus-buy considerations, cloud versus on-premises deployment, technology selection and scalability strategies.
2. AI, Machine Learning and Agent Development
- Design and develop AI agents that automatically gather and consolidate evidence for held manufacturing lots through integration with MES, SPC, genealogy, inline inspection, parametric, wafer-map, test-bin and equipment-history systems.
- Develop a hybrid AI architecture combining traditional machine learning and LLM-based reasoning to support:
- Hold-reason classification.
- Anomaly and process excursion detection.
- Wafer-map pattern recognition.
- Similarity search against historical lot dispositions, MRB cases and 8D reports.
- Evidence-based disposition recommendations.
- Implement agent-based orchestration and tool-calling capabilities to retrieve, correlate and interpret structured and unstructured manufacturing data.
- Develop recommendation mechanisms that provide confidence scores, supporting evidence, source references and clear engineering justifications.
- Ensure AI-generated recommendations are explainable, consistent and verifiable by quality and product engineers.
- Contribute directly to Python development, SQL-based data processing, model development and integration implementation.
3. Engineering Standards and MLOps/LLMOps
- Establish the engineering standards and development toolchain for the AI engineering team, including repository structures, coding standards, branching strategies, code reviews, testing frameworks and CI/CD pipelines.
- Define infrastructure and environment management practices, including Infrastructure as Code (IaC), containerisation and deployment strategies.
- Establish MLOps and LLMOps frameworks covering:
- Experiment tracking and model registry.
- Prompt and agent version management.
- Model and agent evaluation pipelines.
- Performance and drift monitoring.
- Model and prompt rollback mechanisms.
- Cost, latency and resource monitoring.
- Define documentation standards, including HLD/LLD templates, Architecture Decision Records (ADRs), technical specifications and operational runbooks.
- Ensure engineering practices support maintainability, scalability, reliability and long-term system ownership.
4. AI Governance, Quality and Safety Controls
- Design human-in-the-loop mechanisms to ensure appropriate engineering oversight of AI-generated disposition recommendations.
- Define confidence thresholds, approval workflows, escalation paths and override mechanisms.
- Establish a controlled implementation roadmap, progressing from recommendation-only capabilities towards selective automation for validated, low-risk hold categories.
- Implement comprehensive traceability for every AI recommendation, including source data, model versions, prompt versions, retrieved evidence, reasoning outputs and final engineering decisions.
- Ensure architecture and system controls support IATF 16949 requirements and automotive customer audit expectations.
- Work with quality stakeholders to establish evaluation criteria before production deployment, including historical back-testing, false-release rates, escape-risk metrics and recommendation accuracy.
- Implement continuous monitoring, drift detection, exception handling and rollback strategies to minimise operational and quality risks.
5. Technical Leadership and Team Development
- Provide technical leadership and architectural direction to AI engineers throughout the solution development lifecycle.
- Translate HLD and LLD designs into actionable development work packages, technical milestones and implementation plans.
- Conduct architecture walkthroughs and technical knowledge-sharing sessions before development begins.
- Perform design reviews and code reviews to ensure implementation consistency, engineering quality and adherence to architectural standards.
- Work closely with AI engineers to resolve technical challenges and make critical design decisions.
- Mentor and develop engineers, enabling them to take independent ownership of their respective solution components.
- Coordinate with the Integration Architect, Data CoE, Manufacturing IT and site engineering teams to ensure successful end-to-end integration.
6. Deployment, Pilot and Multi-Site Rollout
- Lead the initial implementation and pilot deployment at a selected manufacturing site.
- Work with site stakeholders to validate system functionality, recommendation quality and operational readiness.
- Measure business impact through improvements in lot hold cycle time, engineering effort and disposition efficiency.
- Capture pilot findings and continuously refine the architecture, models and workflows.
- Establish a reusable reference architecture and deployment framework for rollout across semiconductor fabrication, assembly and test manufacturing sites.
Job Requirements
Core Requirements
- Minimum 8 years of experience in AI/ML engineering, solution architecture or related technical roles, including at least 2 years of hands-on experience developing LLM-based or AI agent systems.
- Proven experience architecting and delivering production-grade AI/ML solutions, particularly AI-driven decision-support or intelligent automation systems.
- Strong experience producing HLD and LLD documentation and leading end-to-end enterprise system architecture and integration.
- Demonstrated technical leadership experience in guiding engineering teams from solution design through development, testing and production deployment.
- Strong foundation in traditional machine learning techniques, including:
- Classification and predictive modelling.
- Anomaly and excursion detection.
- Time-series and tabular data modelling.
- Feature engineering.
- Gradient boosting and other supervised learning techniques.
- Model evaluation for imbalanced datasets.
- Hands-on experience with AI agent frameworks, orchestration platforms or SDKs such as LangGraph, AutoGen, Claude Agent SDK/MCP or equivalent custom agent architectures.
- Practical experience implementing Retrieval-Augmented Generation (RAG), tool calling and integration with structured and unstructured data sources.
- Strong programming skills in Python and SQL, with experience handling large-scale tabular, time-series or sensor datasets.
- Experience establishing software engineering practices, including CI/CD, automated testing, code reviews, version control and MLOps.
- Experience designing human-in-the-loop AI systems with strong emphasis on explainability, auditability, governance and fail-safe mechanisms.
- Strong stakeholder management and communication skills, with the ability to work directly with manufacturing, quality and process engineering teams to translate operational requirements into technical solutions.
Good to Have
- Semiconductor manufacturing experience, particularly in front-end wafer fabrication or back-end assembly and test environments.
- Familiarity with manufacturing quality and process engineering concepts, including lot disposition, SPC, yield management, wafer maps, test data and STDF.
- Experience integrating enterprise manufacturing systems such as MES, SPC, yield management, equipment monitoring and Fault Detection and Classification (FDC) platforms.
- Experience with computer vision applications, including Automated Optical Inspection (AOI), Scanning Electron Microscopy (SEM) and visual defect classification.
- Familiarity with architecture modelling and documentation methodologies such as C4, UML and sequence diagrams.
- Experience with enterprise integration patterns, including REST APIs, event-driven architecture, message queues and streaming platforms.
- Hands-on experience with cloud AI platforms such as Azure AI Foundry, AWS Bedrock or SageMaker.
- Experience with Docker, Kubernetes, Infrastructure as Code and enterprise CI/CD platforms.
- Familiarity with MLOps/LLMOps platforms such as MLflow, LangSmith and Arize, including model registries, evaluation frameworks, drift monitoring and cost management.
Skills
- Agentic AI
- AI
- API
- Assembly
- AutoGen
- Automation
- AWS
- AWS Bedrock
- Azure
- CI/CD
- Cloud
- Computer Vision
- Containerization
- Data Ingestion
- Data Modeling
- Docker
- Event Driven Architecture
- Feature Engineering
- Infrastructure as Code
- Kubernetes
- LangGraph
- LangSmith
- LLM
- LLMOps
- Machine Learning
- MCP
- MLflow
- MLOps
- Model Evaluation
- Predictive Modeling
- Python
- RAG
- REST
- SageMaker
- Solution Design
- SQL
- Stakeholder Management
- System Integration
- Test Automation
- Time Series
- UML
- Version Control
Sign in to read and join the discussion.