Generative AI Course using AWS
Module 1: Python (20 hours)
● Introduction to Python:
o Python syntax, data types, and variables
o Functions and modules
o File handling
● Data Structures:
o Lists, tuples, dictionaries, sets
● Object-Oriented Programming:
o Classes and objects
o Inheritance, encapsulation, polymorphism
● Error Handling:
o Exception handling, debugging
● Libraries for AI/ML:
o Introduction to NumPy, pandas, Matplotlib, scikit-learn
● Working with JSON, CSV, and APIs
Module 2: AWS Basics, S3 (10 hours)
● Introduction to AWS:
o Core AWS services (EC2, IAM, RDS)
● Amazon S3:
o Bucket creation and management
o Object storage and access control
o Versioning, lifecycle policies
● Cloud Security:
o IAM roles and policies
o Best practices for securing S3 buckets
Module 3: AWS AI Models, Lambda, Bedrock (10 hours)
● Introduction to AWS AI/ML Services:
o Overview of AWS AI tools (SageMaker, Amazon Q)
● AWS Lambda:
o Serverless architecture, use cases
o Writing Lambda functions in Python
o Event-driven architecture with Lambda
● Amazon Bedrock:
o Overview of Bedrock, foundation models
o Use of pre-trained models on AWS Bedrock
Module 4: AWS Kendra (5 hours)
● Introduction to AWS Kendra:
Generative AI Course using AWS
oIntelligent search services for enterprises
oUse cases: internal documentation, search applications
● Configuring Kendra:
o Data indexing and integration
o Search query optimization
o Security controls and monitoring
Module 5: Flask REST API (10 hours)
● Introduction to Flask:
o Building REST APIs with Flask
o Routing, middleware, request and response
● API Authentication and Security:
o Implementing JWT and OAuth
o Securing APIs using HTTPS, CORS
● Database Integration:
o Connecting Flask with databases
● Deployment on AWS (EC2)
Module 6: LangChain (5 hours)
● Introduction to LangChain:
o Purpose and use cases of LangChain in AI pipelines
● Building with LangChain:
o Creating AI applications with LangChain
o Connecting to external APIs and databases
● Advanced Features:
o Memory management, chains, and agents in LangChain
Module 7: Vector Databases (Chroma DB, FAISS DB) (2 hours)
● Introduction to Vector Databases:
o Understanding vector search and embeddings
o Differences between traditional and vector DBs
● Chroma DB and FAISS DB:
o Installation and setup
o Storing and searching vectors
o Use cases in Generative AI
Module 8: Prompt Engineering (2 hours)
● Introduction to Prompt Engineering:
o Crafting effective prompts for AI models
● Optimizing Prompts:
Generative AI Course using AWS
o Prompt structure, experimenting with variations
o Using prompt engineering to fine-tune responses from models
Module 9: RAG (Retrieval-Augmented Generation) Model (10 hours)
● Overview of RAG Models:
o Combining retrieval-based and generation-based approaches
o Applications in large-scale QA systems
● Implementation with LangChain:
o Integrating with vector DBs and external data sources
● Fine-tuning RAG Models:
o Adapting RAG models for domain-specific tasks
Capstone Project:
"Building a Question-Answering System Using RAG and AWS Services"
● Objective: Build an intelligent question-answering system that leverages AWS AI tools,
vector databases, and a RAG model to provide accurate responses based on large
datasets.
● Steps:
o Set up an AWS S3 for storing documents
o Build a Flask REST API for querying the system
o Use AWS Kendra for document search and retrieval
o Integrate LangChain for managing chains and memory
o Implement a RAG model using FAISS DB or Chroma DB for vector storage
o Create prompts and optimize the model’s output using prompt engineering
o Deploy the final system on AWS
This structure will provide comprehensive hands-on experience for the professionals, making
them proficient in both AWS tools and cutting-edge Generative AI techniques.
