Codebase ingestion
Automated ingestion brings engineering material into a shared knowledge layer rather than leaving context scattered across repositories.
Back to work
Connected knowledge
Connected automated code ingestion with Gmail, Drive and Chat in a unified LLM knowledge system spanning 30+ projects and codebases.
My roleEngineering collaborator
Overview
Engineering context was distributed between codebases and everyday collaboration tools. The project brought those sources together in a unified LLM knowledge system.
Project statusInternal enterprise system
Scope
A collaborative LLM knowledge system bringing code and Google Workspace content together across more than 30 internal projects and codebases.
Main features
Workspace content
Gmail, Drive and Chat contribute the working context around projects, complementing the code itself.
Knowledge authoring & exploration
The web implementation includes Markdown rendering and editing, diagrams and graph-view components for working with connected technical material.
Technology
React
Knowledge interface and content views.
Markdown & CodeMirror
Readable technical content and in-browser editing.
Mermaid
Diagram support in knowledge content.
Python & AWS CDK
Backend application and infrastructure definition.
Lambda & DynamoDB
Processing functions and durable knowledge-system state.
Google Workspace
Source context from Gmail, Drive and Chat.
Engineering approach
Code and conversation provide different evidence
Code explains implementation; project communication explains decisions and working context. Bringing them together makes the system more useful than a repository-only view.
Make knowledge usable after ingestion
Editing, diagrams and navigation support the human side of the workflow. My role was an engineering collaboration within the project team.
What I worked on
- Collaborated with the project team to integrate automated code ingestion.
- Connected Google Workspace content from Gmail, Drive and Chat.
- Helped unify project and codebase context for use through an LLM knowledge system.
Results & scope
30+
internal projects & codebases covered
- Knowledge coverage across more than 30 internal projects and codebases.
- A connected knowledge layer across code and collaboration content.
Built collaboratively within Axrail. Internal source material and code are not published on this page.
Another project
Singpass migration
Contact04
Say hello.
For AI engineering roles, collaborations
or a conversation about further study.