AI System Design

Course Description

Design and architect enterprise-grade AI systems. This advanced intermediate course covers system design patterns, infrastructure choices, cost optimization, security, and scaling strategies for AI-powered applications. You'll design systems that handle real production workloads.

Who Is This For?

• Graduates of Applied AI for Business seeking deeper technical knowledge
• Software engineers and architects adding AI to their stack
• Technical leads responsible for AI infrastructure decisions
• CTOs and technical founders building AI-first products

Prerequisites

• Completion of "Introduction to Generative AI" and "Applied AI for Business"
• Understanding of AI APIs and basic application building
• Some programming experience (Python preferred)
• Degree-level education or equivalent work experience
• Requires application and approval

Syllabus

Module 1: AI System Architecture Fundamentals

• Architectural patterns for AI systems
• Synchronous vs asynchronous AI processing
• Microservices vs monolith for AI applications
• Designing for reliability and fault tolerance

Module 2: LLM Infrastructure

• Self-hosted vs API-based LLM deployment
• Model serving frameworks (vLLM, TGI, Ollama)
• GPU provisioning and cost management
• Multi-model architectures and routing

Module 3: RAG at Scale

• Advanced RAG patterns (hybrid search, re-ranking)
• Vector database selection and scaling
• Chunking strategies for production
• Evaluation frameworks for RAG quality

Module 4: Agent Systems

• Multi-agent architectures
• Tool use and function calling
• Memory and state management for agents
• Safety guardrails and output validation

Module 5: Data Pipeline Architecture

• Real-time vs batch processing
• ETL pipelines for AI data preparation
• Feature stores and embedding management
• Data versioning and lineage

Module 6: Security & Compliance

• Prompt injection prevention
• Data privacy in AI systems (PDPA compliance)
• Audit trails and explainability
• AI governance frameworks

Module 7: Performance & Cost Optimization

• Caching strategies for AI workloads
• Token optimization and prompt compression
• Autoscaling AI infrastructure
• Cost modelling and budgeting

Module 8: Capstone: System Design Review

• Design a complete AI system for a real use case
• Architecture review with industry experts
• Peer design reviews
• Portfolio documentation

What You'll Get

• 30 hours of instructor-led training (10 sessions x 3 hours)
• Architecture template library
• Access to cloud infrastructure for hands-on labs
• 1-on-1 architecture review sessions
• Certificate of completion
• SkillsFuture Credit eligible (pending SSG accreditation)