Skip to content

README.md

Latest commit

 

History

History
118 lines (93 loc) · 3.97 KB

README.md

File metadata and controls

118 lines (93 loc) · 3.97 KB

🚀 8-Week Agentic AI Learning Plan

Goal: Become consulting-ready for Generative AI Solution Architecture with AWS Bedrock, LangChain/LangGraph, RAG, and MCP — with essential Math/ML knowledge for enterprise-grade solutions.

📌 Overview

This program is designed for professionals who are already comfortable with Python and want to build Agentic AI applications ready for enterprise deployment.
The plan blends:

  • Hands-on agentic AI development
  • AWS-native integration
  • Essential math & ML concepts
  • Deployment & consulting skills

📅 Weekly Plan

Week 1 – Core LLM & AWS Bedrock

Goal: Connect to Bedrock, query models, and tune prompts.

  • AWS Bedrock SDK (boto3) setup
  • Calling Claude, Mistral, Titan models
  • Prompt engineering (zero/few-shot, system prompts, CoT)
  • Parameters: temperature, max tokens, stop sequences
  • Math/Stats: Vectors, cosine similarity

Week 2 – Bedrock Embeddings & Knowledge Bases

Goal: Store and query embeddings for contextual AI.

  • Titan embeddings API
  • Bedrock Knowledge Bases basics
  • Embedding search (FAISS)
  • Vector DBs: Pinecone, OpenSearch Vector Engine
  • Math: Dot product, norms, distance metrics

Week 3 – RAG (Retrieval-Augmented Generation)

Goal: Build your first RAG pipeline.

  • LangChain document loaders (PDF, web, CSV)
  • Chunking & metadata tagging
  • Embedding storage & retrieval
  • Context injection into prompts
  • Hallucination detection via retrieval evals

Week 4 – Agents & Tools

Goal: Create an LLM agent that uses tools.

  • LangChain agents (react, zero-shot-react, tool-calling)
  • Built-in tools vs custom tools
  • Memory types (ConversationBuffer, SummaryMemory)
  • AWS S3 tool integration
  • Math: Probability basics & confidence scoring

Week 5 – LangGraph & Multi-Agent Orchestration

Goal: Build stateful workflows for agents.

  • LangGraph basics (nodes, edges, conditional routing)
  • Loops, retries, state persistence
  • Multi-agent orchestration
  • Combining RAG + tools in LangGraph
  • AWS Bedrock Agents

Week 6 – MCP (Model Context Protocol)

Goal: Standardize tool integration for agents.

  • MCP concepts & benefits
  • Running an MCP server locally
  • Exposing APIs/databases as MCP tools
  • Agent consuming MCP tools
  • ML Concepts: Overview of transformers & embeddings

Week 7 – Deployment & Scaling

Goal: Deploy and scale in production.

  • AWS Lambda + API Gateway deployment
  • SAM CLI / CDK for IaC
  • Provisioned concurrency & cold start optimization
  • Step Functions orchestration
  • Monitoring: LangSmith, CloudWatch
  • Security: Secrets Manager, IAM least privilege

Week 8 – Capstone & Consulting Prep

Goal: Build portfolio projects & prepare for client work.

  • Capstone: Multi-agent AWS Bedrock app with RAG + LangGraph + MCP, deployed serverlessly
  • Cost modeling for GenAI workloads
  • Security & compliance checklist (PII, GDPR, HIPAA)
  • Create demo deck & record walkthrough
  • Draft consulting proposal & SoW template

🛠 Tools & Tech Stack

  • Languages: Python (primary), optional Java/Spring Boot for integration
  • LLM Platforms: AWS Bedrock, OpenAI, Hugging Face
  • Frameworks: LangChain, LangGraph, CrewAI, AutoGen
  • Vector DBs: FAISS, Pinecone, OpenSearch Vector Engine
  • Deployment: AWS Lambda, Step Functions, SAM CLI, Docker
  • Monitoring: LangSmith, CloudWatch, OpenTelemetry
  • Math/ML: Numpy, basic linear algebra, probability, transformers overview

🎯 Outcomes After 8 Weeks

  • Design and deploy enterprise-ready Agentic AI solutions
  • Integrate RAG, multi-agent orchestration, and MCP tools
  • Build and deploy on AWS Bedrock with scalability & security
  • Prepare portfolio projects and consulting materials for clients

📂 Suggested Folder Structure for Practice