World-Class Curriculum
Modules
AWS
Infrastructure & Services
Data Science
Processing & Analytics
AI
Development & Agents
Augmentation & MCP
Optimization & Protocol Integration
Weekly Breakdown
Week 1
Setup & Copilot Magic
Real Case: Week 1 Course: Intro for Zero Exp—VS Code + Copilot install; no code yet. We help you set up Git, Copilot, AWS account, Python environment, and a developer desktop. You will write IaC code to create and destroy S3, Lambda, and IAM Role with IaC.
Topics:
- AWS signup, IAM tour via console
- Free Tier explainer (Copilot: 'Explain AWS costs simply')
- Developer desktop setup: Git, Copilot, AWS account, Python
- Hands-on: IaC for S3, Lambda, IAM Role setup
- Ingest data into S3
- Read data from S3 bucket using Lambda
- Data basics: What is CSV? (No code)
Milestone: Copilot setup; run first script
Modules:
AWSData ScienceAIAugmentation & MCP
Week 2
Data Staging Basics
Real Case: Real Case: Kaggle tickets as 'Hyd cab support' (e.g., 'app crash')
Topics:
- Effortlessly upload data to S3 using real code—no manual steps
- Automate ticket processing: Lambda reads directly from S3
- Instantly visualize your data flow—see results as you run code
- Clean messy CSVs in seconds with Copilot-powered scripts
- Quickly plot ticket resolution times for actionable insights
- Instantly filter and analyze 100+ support tickets
- Embed ticket text for advanced search using sentence-transformers
- Discover how agent tools supercharge your data retrieval
- Experience hands-on embedding with real support tickets
- Connect to AWS FAQ with a single Lambda function
- Automatically tag and prioritize your support data
- Summarize ticket batches with AI-powered prompts
Milestone: 100 tickets staged; embedded
Modules:
AWSData ScienceAIAugmentation & MCP
Week 3
RAG Pipeline Starter
Real Case: Real Case: Chunk like 'Ola ride issues' from Hugging Face
Topics:
- OpenSearch create (console wizard)
- Copilot: 'Ingest to OpenSearch from S3'
- Hands-on: Run script; check index
- Chunking easy: Copilot: 'Split ticket text by sentences'
- Supercharge your data enrichment: Embed all chunks automatically with Lambda
- Achieve clean, high-quality data: Remove duplicates and noise using Lambda audit
- Unlock best-in-class embedding: Compare file-based vs Lambda-powered approaches
- Empower instant answers: Enable semantic search with a custom Lambda function
- Accelerate AI adoption: Bedrock models, roles, and policies made simple
- OpenSearch create (console wizard)
- Copilot: 'Ingest to OpenSearch from S3'
- Hands-on: Run script; check index
- Chunking easy: Copilot: 'Split ticket text by sentences'
- Hands-on: Check output (no math)
- RAG flow: Copilot: 'LangChain RAG chain template'
- LLM call: 'Query with Hugging Face'
- Hands-on: 'Outage' → chunk retrieval
- MCP ground: Copilot: 'Add AWS context to prompt'
- Threshold: 'If low score, alert'
- Hands-on: Augment one response
Milestone: 300 chunks embedded, deduplicated, and ready for semantic search with Bedrock integration
Modules:
AWSData ScienceAIAugmentation & MCP
Week 4
Retrieval & Simple Gen
Real Case: Real Case: DoorDash 'delay fix' adapted to 'Hyd traffic tickets'
Topics:
- Unlock instant answers: Lightning-fast Lambda search with k-NN queries
- Gain full visibility: Add CloudWatch logging for every retrieval
- Experience real-time results: Test async runs and see immediate feedback
- Boost relevance: Rerank results by keywords for smarter answers
- Measure what matters: Check recall and accuracy with easy metrics
- Spot and fix issues fast: Hands-on troubleshooting with real samples
- Get grounded, reliable responses: Use Copilot prompts for resolution
- Achieve next-level accuracy: Agent-powered retrieval and generation
- See Copilot in action: Output for real support tickets
- Tap into Bedrock: Fetch prompts for advanced AI workflows
- Accelerate with speed: Use small models for quick, cost-effective results
- Reduce hallucinations: Fact-check responses with MCP tools
- Master rewriters, rerankers, cosine similarity, and Euclidean distance for ultra-accurate, trustworthy answers
Milestone: 3-query test
Modules:
AWSData ScienceAIAugmentation & MCP
Week 5
UI & API Intro
Real Case: Real Case: Flipkart 'order query' chat
Topics:
- Launch powerful APIs: Instantly set up REST endpoints with API Gateway
- Secure your app: Add authentication in minutes with Cognito
- Build modern UIs: Create interactive apps with Streamlit
- Capture feedback effortlessly: Log user actions to CSV with Copilot
- Visualize insights: Generate bar charts for instant query analysis
- Compare and optimize: Test multiple UI versions for best results
- Empower users: Add smart RAG input fields with Streamlit agents
- Keep conversations contextual: Remember chat history for seamless support
- Demo real-world help: Provide instant EC2 assistance via UI
- Boost support: Add AWS help buttons for one-click guidance
- Measure quality easily: Check scores with simple built-in tools
- Integrate AI responses: Link MCP outputs directly in your UI
Milestone: Local UI and API live—query Lambda, evaluate RAG, and deliver instant support
Modules:
AWSData ScienceAIAugmentation & MCP
Week 6
Deploy & Interact
Real Case: Real Case: Zomato 'food delay' bot deploy
Topics:
- Deploy with confidence: Launch your solution using Lambda—pay only when you use it
- Optimize user experience: Set up throttling for smooth scaling
- Store and rate instantly: Use DynamoDB for real-time feedback
- Visualize success: Evaluate results using interactive Streamlit app dashboards
- Analyze conversations: Dive into 10 real chat sessions for actionable learnings
- Deliver seamless support: Enable multi-turn follow-ups with smart agents
- Integrate full RAG: Power your Streamlit app with advanced retrieval and generation
- Simulate real tickets: Hands-on practice for real-world scenarios
- Expose live insights: Use MCP API for instant analytics
- Measure speed: Track and optimize response times for peak performance
- Get smart UI suggestions: Enhance your interface with MCP-driven recommendations
- Set up FAQ routers, personalize user journeys, reduce latency, and validate performance for a flawless rollout
Milestone: Production-ready deployment with personalized support, low latency, and validated performance—ready for real users
Modules:
AWSData ScienceAIAugmentation & MCP
Week 7
MCP Server & Client Intro
Real Case: Real Case: Set up a sample MCP server and client for cybersecurity alert tickets (Hyd/Anduril use case)
Topics:
- Get started: MCP Server and Client architecture overview
- Hands-on: Spin up a sample MCP server for ticket processing
- Connect your client: MCP Client setup and integration
- Cache for speed: Add Redis to MCP workflows
- Scale easily: Lambda auto-config for MCP endpoints
- Monitor costs: Use Cost Explorer for MCP deployments
- Fine-tune for your data: LoRA intro on small ticket sets
- Spot errors fast: Hallucination check scripts for MCP outputs
- Personalize responses: Add custom logic to MCP Client
- Secure your pipeline: Best practices for MCP server/client security
- Analyze and optimize: Review performance and accuracy
- Deploy and test: End-to-end MCP workflow simulation
- Hands-on: Latency graph
- Hybrid search: Copilot: 'Keywords + vectors'
- MCP action: 'Multi-tool agent'
- Hands-on: Route secure queries
- Custom MCP: 'Build one tool prompt'
- Scores: 'Add confidence'
- Hands-on: Bias flag in outputs
Milestone: Sample MCP server and client deployed, integrated, and optimized for real-world ticket processing
Modules:
MCP ServerMCP Client
Week 8
Capstone & Polish
Real Case: Real Case: Build a mentor query bot for T-Hub startup using agentic AI and MCP RAG agent
Topics:
- Save costs: Automate cleanup and teardown for efficient resource use
- Benchmark your solution: Compare baselines with Copilot-powered evals
- Showcase results: Visualize your portfolio with interactive graphs
- Scale up: Run 200+ samples for robust testing
- Empower teams: Build multi-user agentic AI systems with Copilot
- Build ethical AI: Address bias in local data for fair outcomes
- Polish your project: Apply final tweaks for a professional finish
- Add advanced search: Integrate MCP web tools for powerful queries
- Customize for impact: Augment with Hyd-specific features (e.g., weather)
Milestone: Team-built agentic AI system using MCP and RAG agent, with polished UI, metrics, and T-Hub pitch-ready demo
Modules:
MCPAI RAG AgentTeamwork