Josh396s/email.ing
A privacy-first, intelligent email management system. Utilizes a hybrid AI architecture with local Llama models for secure classification and Google Gemini for context-aware summarization and drafting.
What's novel
A privacy-first, intelligent email management system. Utilizes a hybrid AI architecture with local Llama models for secure classification and Google Gemini for context-aware summarization and drafting.
Code Analysis
5 files read · 2 roundsA privacy-first email management tool that locally classifies emails using Llama 3 and summarizes them using Gemini 2.5 Flash after masking PII.
Strengths
The project implements a sophisticated hybrid AI architecture (Local Llama + Cloud Gemini) with a robust PII masking pipeline using Presidio. It effectively separates concerns between ingestion, AI processing, and storage, ensuring sensitive data is encrypted at rest and masked before leaving the local environment.
Weaknesses
The code lacks unit tests entirely. Error handling in the AI service is somewhat brittle (e.g., regex-based JSON parsing for Ollama). The README mentions a specific Gemini model version (`gemini-2.5-flash`) which may not exist yet, potentially causing runtime errors.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
49
Contributors
1
Files
55
Active weeks
11
Repository
Language
Python
Stars
1
Forks
0
License
—