Skip to content

Add chatbot_development course to technology_and_tooling#260

Open
alokkrsahu wants to merge 3 commits into
UNIVERSE-HPC:mainfrom
OxfordCompetencyCenters:chatbot-development
Open

Add chatbot_development course to technology_and_tooling#260
alokkrsahu wants to merge 3 commits into
UNIVERSE-HPC:mainfrom
OxfordCompetencyCenters:chatbot-development

Conversation

@alokkrsahu

Copy link
Copy Markdown

New applied AI course authored by the Oxford AI Competency Centre. Content covers:

  • Module 1 (4 lessons): Chatbot foundations and OpenAI API integration
  • Module 2 (5 lessons): RAG with vector embeddings and ChromaDB
  • Module 3 (5 lessons): Advanced RAG - agentic retrieval, reranking, hybrid search, knowledge graphs, semantic caching, evaluation
  • Module 4 (5 lessons): Memory and context management including auto-summarisation and retrieval-based memory
  • Modules 5-7 (15 lessons, tagged 'optional'): Deployment, analytics, and production-hardening topics for self-paced study
  • Module 8 (2 lessons): Streamlit-based capstone project brief and deliverables/documentation guide

Frontmatter uses the standard course-material schema (name, id, dependsOn with dotted paths, tags, learningOutcomes, attribution). Exercises are wrapped in :::challenge directives where the source contained explicit exercise sections. Runnable notebooks and the Streamlit capstone scaffold live in the companion repository at https://github.com/OxfordCompetencyCenters/aicc_chatbot_modules and are linked from each lesson's "Hands-on practical" section.

Also adds chatbot_development to the courses list in technology_and_tooling/index.md.

New applied AI course authored by the Oxford AI Competency Centre.
Content covers:

- Module 1 (4 lessons): Chatbot foundations and OpenAI API integration
- Module 2 (5 lessons): RAG with vector embeddings and ChromaDB
- Module 3 (5 lessons): Advanced RAG - agentic retrieval, reranking,
  hybrid search, knowledge graphs, semantic caching, evaluation
- Module 4 (5 lessons): Memory and context management including
  auto-summarisation and retrieval-based memory
- Modules 5-7 (15 lessons, tagged 'optional'): Deployment, analytics,
  and production-hardening topics for self-paced study
- Module 8 (2 lessons): Streamlit-based capstone project brief and
  deliverables/documentation guide

Frontmatter uses the standard course-material schema (name, id,
dependsOn with dotted paths, tags, learningOutcomes, attribution).
Exercises are wrapped in `:::challenge` directives where the source
contained explicit exercise sections. Runnable notebooks and the
Streamlit capstone scaffold live in the companion repository at
https://github.com/OxfordCompetencyCenters/aicc_chatbot_modules and
are linked from each lesson's "Hands-on practical" section.

Also adds chatbot_development to the courses list in
technology_and_tooling/index.md.
Gutenberg's AttributionDialog component (components/dialogs/
AttributionDialog.tsx) expects every citation to carry an `image`
field: it renders `<Image src={citation.image} ... />` on line 70 and
calls `citation.image.startsWith("./")` on line 95. Our lessons
originally omitted the field, which crashed the dialog with
"Cannot read properties of undefined (reading 'startsWith')" when
visited. Point every citation.image at the OxfordCompetencyCenters
GitHub org avatar, matching the pattern used by the upstream docker
course.
Previously every lesson depended on exactly the one immediately before
it, producing a 37-node linear chain in Gutenberg's DAG-view diagram.
Restructure so each module becomes a star: later lessons within a
module depend on that module's first ("hub") lesson, and module hubs
connect to the last lesson of the previous module via a single bridge
edge.

The diagram now renders as one hub per module with its sibling lessons
radiating out, and clear bridge edges between modules. Navigation via
the sidebar is unaffected; sequential reading order is preserved.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant