Che Ndandala (Personal AI Assistant)

Meet Che Ndandala, a lightweight AI personal assistant built on top of open source and openly available AI models. You’ll find Che embedded in the bottom right corner of this website, ready to answer questions about my work, projects, technical experience, and the ideas documented throughout my portfolio but also some simple tasks

Think of it as an interactive layer on top of my portfolio. If you don't have time to browse through every page, just ask Che Ndandala.

Rather than simply presenting a long list of blog posts,skills, projects, and experiences, I wanted to build a digital representation of my work that you could actually have a conversation with. You can ask something specific about one of my projects, my engineering experience, my medical background, or the technologies I've worked with, and Che will search through the knowledge I've provided and formulate an answer for you.

The assistant uses a Retrieval-Augmented Generation (RAG) approach, combined with other agentic workflows.

Che can also act as a small bridge between a visitor and me.

If someone is genuinely interested in contacting me, Che can collect their name, email address, and the nature of their interest and forward that lead to me.

So a conversation can naturally go from:

"What technologies does Seba use?"

to:

"I'd like to work with Seba on a project. My email is..."

It can also help visitors navigate the website.

If you ask:
"Take me to Seba's projects or blog posts."
or
"Show me his GitHub."
the assistant can trigger a navigation action instead of simply replying with a URL.This is a small feature, but it demonstrates an important idea that assistant doesn't have to only generate text. It can perform actions

Why build it?

Because I wanted to experiment with something that sits in conjunction with software engineering, AI, and communication.

Building Che Ndandala gave me the opportunity to work with technologies such as RAG, vector search, embeddings, LangGraph, LLM tool calling, and API-based AI systems, while also solving a practical problem: making a large and diverse portfolio easier to explore.

And there's something satisfying about turning a portfolio from something you read into something you can talk to.

So, instead of scrolling through pages trying to piece together what I do, just ask Che, when you have time, reading the full page is important!

Go ahead -> the chat bubble is waiting in the bottom-right corner