Home / Projects
Learn by Building Real Projects
Don’t just watch tutorials. Build projects that demonstrate what you can actually do. Each brief below is part of the GenAI & Agentic AI Builder, which ends with a capstone based on an OptnGrow project brief.
Python / LLM API / LangChain
Study Assistant
Summarise study notes and generate practice questions with an LLM, using prompt templates and structured output.
LLMs / Embeddings / RAG
Document Q&A Assistant
Answer questions from a set of supplied documents, show the source passage, and say so when the answer is not there.
Agentic AI / Tool Calling
Task Assistant Agent
An assistant that decides when to call a predefined Python function, uses the result and responds within clear limits.
What each project involves
The problem, what you build, and what you walk away with.
GenAI & Agentic AI Builder
Study Assistant
Summarise study notes and generate practice questions with an LLM, using prompt templates and structured output.
Problem
Learners have pages of notes but little time to review them or test themselves.
What you build
A Python and LangChain app that summarises notes with prompt templates and generates practice questions in a consistent, structured format.
Deliverables
Working app, prompt comparison notes and a README.
Tools
Python / LLM API / LangChain
GenAI & Agentic AI Builder
Document Q&A Assistant
Answer questions from a set of supplied documents, show the source passage, and say so when the answer is not there.
Problem
Answers to common questions are buried in FAQs, course material or policy documents.
What you build
Chunking, embeddings and a vector store to retrieve relevant text, plus an LLM that answers only from that context and says when the documents do not cover the question.
Deliverables
Q&A app that shows source passages, a test log of normal and unanswerable questions, and a README.
Tools
LLMs / Embeddings / RAG
GenAI & Agentic AI Builder
Task Assistant Agent
An assistant that decides when to call a predefined Python function, uses the result and responds within clear limits.
Problem
Some requests need more than a text answer, such as a calculation or a lookup in supplied information.
What you build
A controlled workflow where the LLM chooses a predefined Python function as a tool, passes the result back and responds, with limits, error handling and human review.
Deliverables
Working agent demo, tested scenarios and known limitations, and a README.
Tools
Agentic AI / Tool Calling
Learner showcases are coming soon
As our learners complete their capstones, we’ll feature their projects here with links to their GitHub repositories and live demos.
Want to build projects like these?
Talk to a mentor about the program and the projects you’ll build.