Course Details
Building Intelligent Systems with RAG and Large Language Models (LLMs)
This 5-day hands-on workshop teaches participants how to design, build, and deploy intelligent systems using Retrieval-Augmented Generation (RAG) and Large Language Models (LLMs). Participants will learn to ingest and index organizational data, build retrieval pipelines, integrate LLMs for context-aware responses, and deploy AI applications in real-world scenarios. The training emphasizes practical application, including automating knowledge retrieval, supporting decision-making, and enhancing customer engagement. Through guided exercises and project-based learning, participants will acquire skills to build scalable AI systems that leverage organizational knowledge and LLMs effectively. By the end of the workshop, attendees will have hands-on experience creating intelligent systems ready for immediate implementation, enabling their organizations to innovate, optimize processes, and derive actionable insights from data with AI.
Learning Outcomes:
Course Outline:
Overview of Large Language Models (LLMs) and Generative AI
Understanding Retrieval-Augmented Generation (RAG)
Applications in enterprise and knowledge management
Responsible AI, ethics, and bias mitigation
Customer support automation
Knowledge retrieval and internal wiki automation
Decision support systems
Case studies of RAG + LLM in industry
Structured vs unstructured data
Data cleaning and normalization
Text vectorization and embeddings
Handling multilingual or domain-specific data
Loading sample datasets
Creating embeddings for RAG
Preparing documents for retrieval pipelines
Analyze raw data and preprocess it for AI ingestion
Generate initial embeddings using open-source libraries
Pipeline architectures and workflows
Query understanding and retrieval logic
Ranking and filtering results
Introduction to vector databases (FAISS, Pinecone, Weaviate, Milvus)
Indexing and querying embeddings
Similarity metrics and optimization
Hybrid search: combining keyword and semantic search
Relevance tuning and query expansion
Handling large datasets efficiently
Metrics for accuracy and relevance (precision, recall, MRR)
Testing pipelines with example queries
Improving performance through feedback loops
Build a full retrieval pipeline
Index multiple datasets and perform semantic search
Evaluate pipeline performance
Connecting RAG pipelines with LLM APIs (OpenAI, LLaMA, Anthropic, etc.)
Request-response patterns
Handling token limits and API optimization
Effective prompt design for context-aware generation
Few-shot prompting techniques
Handling ambiguous or incomplete queries
Domain adaptation and embeddings tuning
Custom model training for specific knowledge domains
Monitoring model outputs for quality
Designing question-answering systems
Automating FAQs and internal knowledge retrieval
Hybrid human-AI workflows
Integrate LLM with retrieval pipeline for QA system
Experiment with prompts and evaluate outputs
Customize responses for domain-specific context
Hosting AI systems on cloud platforms (AWS, Azure, GCP)
Using containers (Docker, Kubernetes) for AI services
Monitoring and logging AI systems
Trigger-based pipelines (emails, chatbots, internal apps)
Integrating AI workflows with Slack, Teams, or web portals
Scheduled updates and automated knowledge indexing
Data protection and privacy regulations in Malaysia
Secure handling of sensitive data in AI systems
Authentication and API security best practices
Optimizing embeddings and retrieval speed
Scaling systems for large enterprise datasets
Load testing and resource management
Deploy a QA chatbot to the cloud
Automate retrieval of new knowledge documents
Test performance and optimize pipeline
Define a real-world use case (knowledge management, customer support, or decision support)
Implement RAG + LLM end-to-end pipeline
Integrate retrieval, LLM generation, and deployment
Multi-turn conversation and context tracking
Dynamic knowledge updates
Personalized responses and adaptive AI workflows
User satisfaction metrics
Latency, throughput, and system reliability
Continuous improvement and retraining strategies
Large multimodal models (text, image, audio)
RAG for enterprise search, chatbots, and automated insights
AI innovation roadmap and strategic planning
Complete capstone project and present to instructors
Test and optimize AI system with sample queries
Explore optional advanced features (multi-turn conversation, personalization)
Review and discussion of tools and concepts covered.
Q&A session to address any questions.
Let’s Upskill Teams Together.
