Home/Courses/AI Builder/AI Builder: Advanced RAG & Agentic Systems
Intermediate CoursePart of AI Builder

AI Builder: Advanced RAG & Agentic Systems

Build retrieval-augmented and agentic AI systems end-to-end. Over five weeks, implement standard and advanced RAG pipelines, tool-calling agents, graph-based multi-agent workflows, and autonomous systems — culminating in a fully autonomous research agent.

No rating yet
5 weeks

What You'll Learn

Build a complete RAG pipeline from embedding to response generation
Implement hybrid search combining dense and sparse retrieval with cross-encoder re-ranking
Define tool schemas with Pydantic and trigger external APIs autonomously
Orchestrate multi-agent workflows with CrewAI and LangGraph cyclic state machines
Add Human-in-the-Loop breakpoints and agent self-correction loops

Course Content

W1
Week 1: Vector Databases & Standard RAG
Build the foundational retrieval pipeline every production AI system needs.
1
Vector DB Architectures
Comparing Pinecone and Qdrant architectures — indexing strategies, filtering capabilities, and scaling trade-offs.
2
Generating Embeddings & Indexing
Embedding document chunks and building a searchable vector index with metadata for filtered retrieval.
3
Retrieval & Nearest-Neighbor Lookups
Executing approximate nearest-neighbor queries and tuning the top-k parameter for precision vs. recall.
4
Prompt Augmentation & Token Budgeting
Injecting retrieved context into prompts while staying within token limits using dynamic truncation strategies.
Weekly Win
Response Generation & Evaluation
Build a full RAG pipeline — ingest, embed, retrieve, augment, generate — and evaluate end-to-end answer quality.
W2
Week 2: Advanced Retrieval & Re-ranking
Retrieve the right context every time by combining multiple retrieval signals.
1
Keyword vs. Semantic Retrieval
Understanding where BM25 keyword search outperforms dense semantic retrieval and vice versa.
2
Hybrid Search
Combining dense vector and sparse BM25 retrieval with Reciprocal Rank Fusion for superior recall.
3
Metadata Pre-filtering
Restricting the search space with metadata filters before vector lookup to improve precision and reduce cost.
4
Cross-Encoder Re-ranking
Applying a Cross-Encoder model to re-score and reorder the top retrieval candidates for maximum relevance.
Weekly Win
Semantic Routing
Build a semantic router that classifies a query's intent and directs it to the most relevant specialized vector database.
W3
Week 3: Tool Calling & Basic Agents
Give your AI the ability to act, not just respond.
1
Autonomous Decision-Making vs. Scripted Workflows
When to use deterministic function chaining versus letting an agent decide its own execution path.
2
Defining Tool Schemas with Pydantic
Writing typed, validated tool definitions that an LLM can reliably call with correctly structured arguments.
3
Triggering External APIs Autonomously
Wiring tool schemas to live API endpoints so the agent can fetch data, post updates, and trigger workflows.
4
Integrating SQL Database Queries
Exposing SQL queries as agent tools so the LLM can retrieve structured business data on demand.
Weekly Win
OODA Loop Agent
Design and implement an OODA (Observe–Orient–Decide–Act) loop agent that autonomously solves a multi-step research task.
W4
Week 4: Graph-Based Workflows
Orchestrate teams of specialized agents working in parallel.
1
Multi-Agent Paradigms & Worker Networks
Architectures for supervisor/worker agent networks where a planner delegates to specialized executor agents.
2
CrewAI for Role-Based Delegation
Defining agent roles, goals, and backstories in CrewAI so each agent specializes in a distinct task domain.
3
LangGraph for Cyclic State Machines
Building stateful, cyclic agent workflows in LangGraph where agents can loop, branch, and revisit prior steps.
4
Memory & Global Workflow State
Managing shared state across agents using LangGraph's global state object and agent-scoped memory stores.
Weekly Win
Human-in-the-Loop Breakpoints
Add HitL breakpoints to a LangGraph workflow so a human reviewer can approve or redirect agent decisions at critical steps.
W5
Week 5: Autonomous Systems & Capstone
Build agents that run in the background, reflect, and self-correct.
1
Ambient Agents & Background Processes
Designing agents that run on schedules or event triggers without requiring a human to initiate each run.
2
Agent Reflection & Self-Correction
Implementing reflection loops where the agent critiques its own output and reruns steps that fail a quality threshold.
3
Parallelizing Tool Execution
Running multiple tool calls simultaneously using async execution to reduce end-to-end latency at scale.
4
Secure Tool Integration
Safely wiring Qdrant, SQL databases, and web search tools with scoped credentials and input sanitization.
Weekly Win
Capstone: Autonomous Research Agent
Deploy an autonomous research agent with hybrid RAG, parallel tool execution, reflection loops, and HitL checkpoints.

Prerequisites

Python programming
Basic AI understanding
Familiarity with APIs
📚
Intermediate Level
Course Price
₹9,999
India
$199
International · One-time payment
Next cohort starts Mar 30
Duration5 weeks
LevelIntermediate
FormatCohort-based
Modules5

What's included:

Live cohort sessions
Hands-on projects
Certificate of completion
Lifetime access
Career support

Part of Learning Track

🛠️
AI Builder
6 courses in track