Three real AI products, built end to end.
In the Generative AI program I teach at Skillancy, every learner builds, evaluates and deploys these three products, from LLM fundamentals to a live portfolio. The same patterns power my corporate bootcamps.
In the Generative AI program Rishav Kumar teaches, learners build and deploy three AI products: RetailGenie AI (a RAG-based retail assistant), InfoNavigator AI (semantic search with embeddings and a vector database) and DocuMind AI (document understanding with LangChain and responsible-AI checks).
The three products
RetailGenie AI
Intelligent Retail Assistant
A conversational product-recommendation assistant that uses LLMs and retrieval-augmented generation to help shoppers find exactly what they need.
See how it's built →02InfoNavigator AI
Smart Search & Navigation System
A semantic search system built on vector databases and embeddings, retrieving information by meaning instead of keywords.
See how it's built →03DocuMind AI
Intelligent Document Understanding Engine
A document parsing and understanding pipeline built with LangChain, with responsible-AI practices built in throughout.
See how it's built →RetailGenie AI
Intelligent Retail Assistant
A conversational product-recommendation assistant that uses LLMs and retrieval-augmented generation to help shoppers find exactly what they need.
- Prompt engineering: zero-shot, few-shot, chain-of-thought and structured outputs
- LangChain chains and memory for multi-turn conversations
- Agent logic with the ReAct pattern and tool-calling
- RAG grounding on a product catalogue using a vector database
- An image-generation feature and a fine-tuned component (LoRA/PEFT)
Deployed as a live, shareable app (FastAPI / Streamlit) and presented as part of the learner's AI portfolio.
DocuMind AI
Intelligent Document Understanding Engine
A document parsing and understanding pipeline built with LangChain, with responsible-AI practices built in throughout.
- Document parsing and understanding pipeline in LangChain
- Hallucination detection and retrieval-quality checks
- LLM-as-judge evaluation
- Defences against prompt injection, plus content moderation and data privacy
Deployed as a live, shareable app (FastAPI / Streamlit) and presented as part of the learner's AI portfolio.
300+ professionals, one hour, a working AI agent each.
Working professionals had put off learning AI because it seemed too technical.
A live, hands-on Skillancy session: one guided build, step by step, with no setup barriers.
Every one of the 300+ attendees left with a working AI agent they built themselves in 60 minutes.
Tools learners use
Projects FAQ
What AI projects do learners build with Rishav Kumar?
Learners build and deploy three end-to-end Generative AI products: RetailGenie AI, a RAG-based conversational retail assistant; InfoNavigator AI, a semantic search and navigation system; and DocuMind AI, a document understanding engine built with LangChain.
How is RAG used in these projects?
Retrieval-augmented generation (RAG) fetches relevant information from a knowledge source and passes it to the LLM so answers are grounded in real data. RetailGenie AI uses RAG over a product catalogue to recommend products, and InfoNavigator AI uses embeddings and a vector database to retrieve information by meaning.
Which tools are used to build the projects?
Python, OpenAI, LangChain, vector databases for RAG, Hugging Face, PyTorch, TensorFlow, Scikit-learn and Google Colab, with apps deployed using FastAPI or Streamlit.
Can a company's team build similar AI projects with Rishav Kumar?
Yes. The Team Bootcamp and Multi-week Program for enterprises use the same build-evaluate-deploy approach, applied to the company's own use cases and data.
Want your team to build products like these?
Hands-on LLM, RAG and agent bootcamps where your team builds on your own use cases.