Final-year academic project / 02
Mini Chirps
A website-content assistant that retrieves relevant passages before generating an answer. An exploration of local AI and useful context.
Software Engineering project
AI / ML / Retrieval / Full stack
The question
How can an assistant answer questions using a website’s own content? Mini Chirps was my final-year project exploring that question, alongside a lead-capture flow.
My contribution
I made Mini Chirps as an academic software engineering project. The public source brings together a Next.js and TypeScript interface, a Python FastAPI backend and an Ollama-based local language model workflow.
From page to context
The pipeline crawls website text, splits it into passages and generates embeddings with sentence-transformers. The retrieval code compares stored embeddings with the question using cosine similarity and selects the most relevant passages for context.
Scope and evidence
The repository is the best place to inspect the implementation. This is a final-year project; no commercial adoption, benchmark results or production reliability claims are implied.
NEXT CONNECTION
Let’s build
something useful.
I’m seeking a software engineering role. If your team cares about thoughtful systems and the details that make them work, I’d like to hear from you.