Step 2 of 3 · Build · Intermediate

Build AI Applications: RAG & Agentic Workflows

Two days, one integrated industry project. You design a document knowledge assistant that answers with citations, then extend it into an agentic workflow that searches, calls APIs and asks for human approval, tested against realistic and adversarial cases.

Reserve my seat See the curriculum

12 hours live8 hands-on labsIndustry use caseLive & mentor-guided

Duration2 days · 12 hours live

FormatLive online · hands-on

Fees & datesShared on WhatsApp

Reserve my seat

Not sure yet? Join the free 90-min webinar

The industry use case

One real problem, start to finish

Every concept in this workshop is applied to the same realistic scenario, so you learn how the pieces fit together, not just what they are.

The company

A 2,000-employee company with HR and IT policies spread across hundreds of PDFs

The problem

Employees wait days for answers that are already written down; HR and IT teams answer the same questions every week.

What you build

A policy assistant that answers with citations and says "I don't know" when it should, plus an agent that raises IT tickets and leave requests through an API, with a human approval step.

Who it's for

Is this workshop for you?

Developers who finished ExploreThe natural next step: one complete, portfolio-ready application.

Backend & full-stack developersShip RAG and agent features into real products.

Data & ML engineersMove from models and pipelines into LLM application engineering.

Final-year students who code confidentlyA substantial project that stands out in interviews.

Before you join

Helpful: the Explore workshop or some experience calling an LLM (not required)
A laptop and a stable internet connection

Open entry: anyone can join. Prefer a gentler start? Try Explore.

This is not

Five disconnected mini-demos
A no-code chatbot builder
Slides without building

Curriculum

2 days · 10 sessions · 8 hands-on labs

Day 1

Design and build a production-minded RAG application

Design first
- Use-case canvas: users, questions, data sources, success criteria
- Deciding what "good" looks like before writing code
Ingestion that holds upLab
- Parsing PDFs and web pages, cleaning and metadata
- Chunking strategies and what actually breaks retrieval
Embeddings and vector searchLab
- Vector databases, metadata filters and hybrid search
- Reranking concepts and when they are worth it
Grounded answersLab
- Citations, source-grounded prompts and refusing to guess
- A Q&A API with a simple web interface
Evaluate itLab
- A 25-question test set from real policy questions
- Retrieval hit-rate, groundedness and hallucination review

By the end of Day 1: Policy knowledge assistant with citations, a web UI and an evaluation report.

Day 2

Implement an agentic workflow on top

LangGraph in depthLab
- State, nodes, edges and conditional routing
- Using your RAG system as a tool
Agents that actLab
- Calling a ticketing API and an employee-data API (mocked)
- Preparing structured requests the business can process
Reliability by designLab
- Validation, bounded retries and fallbacks
- Human approval before anything irreversible
Testing like an engineerLab
- Realistic scenarios, edge cases and adversarial prompts
- Fixing what the tests reveal
Package and demo
- Docker Compose, README and architecture diagram
- 5-minute demo day with feedback

By the end of Day 2: Integrated RAG + agentic workflow application, packaged with Docker and demoed.

Tools & stack

PythonLangGraphVector DB (Chroma / pgvector)EmbeddingsFastAPIStreamlitDocker

Inside the labs

See what a lab looks like

Real recordings from our hands-on labs: code, running apps and measured results.

RAG lab

A documentation assistant, built end to end

A lab project: FastAPI backend, Chroma vector store and Llama 3 via Groq, answering questions about LangChain documentation.

RAGFastAPIChromaLlama 3

RAG evaluation lab

Measuring answer quality, not guessing

A lab notebook: a local FAISS knowledge base, a RAG pipeline on Groq LLMs, and RAGAS metrics to evaluate the answers. Shown at 2.5× speed.

RAG evaluationRAGASFAISSGroq

Outcomes

After two days, you will be able to…

Design an AI application from a business problem, with clear success criteria
Build a RAG system that cites sources and knows when not to answer
Measure retrieval and answer quality instead of guessing
Implement an agentic workflow with routing, tools, retries and human approval
Package and present an AI application others can run

You leave with

Portfolio-ready GitHub repository (RAG + agent)

Evaluation report and architecture diagram

Course materials & lab notebooks

Verifiable certificate of completion

Next stepStep 3 · Enterprise AI Engineering: From Prototype to ProductionMake your application production-ready →

Your mentor

Taught live by a practising AI engineer

Dr. Gajanan

Lead Mentor · Machine Learning, Generative AI & Agentic AI

14+years in ML & AI

ML→GenAI→Agentsthe full arc of modern AI

Designs the curriculum and leads the hands-on labs, helping participants think through problems the way an AI engineer does instead of memorising tools.

Machine learningGenerative AIAgentic AI systemsPython for AI

FAQ

Questions about this workshop

Do I have to take Explore first?

No. Anyone can join any workshop. Explore makes Build easier, but it is not required.

Which vector database do we use?

Chroma for the labs, with pgvector shown for production. The concepts transfer to Pinecone, Weaviate and others.

Can I use my own documents?

Yes. We provide a realistic policy dataset, and you are welcome to bring non-confidential documents of your own.

Will this help in interviews?

You leave with one integrated, documented project and an evaluation report, which is far stronger than a list of tutorials.

Reserve your seat

Build: Build AI Applications

Share your details. Cohort dates and a secure Razorpay payment link arrive on WhatsApp. Nothing is charged until you confirm.

Fees & datesOn WhatsApp, within minutes

Duration2 days · 12 hours

Full refund if you cancel 48+ hours before the start
Course materials and a verifiable, LinkedIn-ready certificate
Planning more than one workshop? Talk to us

Reserve my seat

Takes under a minute.

Two days to your first complete AI application.

2 days · live online · hands-on

Reserve my seat Try the free webinar first