I fed 28 years of my life into a local server (Python + Local RAG)
Автор: Botwork
Загружено: 2026-01-14
Просмотров: 54
I spent early 2025 building "Project Trace"—a personal semantic memory engine running entirely on my local NAS.
I fed it 28 years of my digital life: 10,000+ journals, emails, notes, and photos from 1997 to today. The goal wasn't to build a product, but to see if I could use local AI to make sense of the patterns in my own history.
This video is a raw demo of the system running on localhost, the architecture behind it, and why I built it.
The Tech Stack:
Language: Python
Search: Dual FAISS Indexes (Semantic + Contextual)
Inference: Qwen 2.5 (32b) via Ollama
Privacy: Custom PII redaction pipeline
Frontend: Simple Flask/React UI
Read the full Engineering Docs: https://botwork.com/trace/
Timestamps: 0:00 - The Context Gap (Why I built this) 2:15 - The "Digital Hoarding" Problem 4:30 - Demo: Querying 28 years of data 6:50 - System Architecture (The 5-Stage Pipeline) 7:15 - Conclusion
Newsletter: / demo-i-fed-28-years-my-life-local-rag-pipe...
Доступные форматы для скачивания:
Скачать видео mp4
-
Информация по загрузке:
