← All work
ARCHIVE · ENTRY 05 · CLASS: PUBLIC
Active Prototype2025–present
AstraOS
Memory-augmented AI workspace
A privacy-first personal AI workspace with persistent memory and document-grounded retrieval.
how it resolves
Query
Retrieval
Memory
Local LLM
Response
Problem
General assistants forget everything and send everything to the cloud. What does a private, persistent, local-first AI workspace look like as a real product?
Built
A FastAPI + React system with local LLM workflows, streaming chat, persistent memory, and document-grounded retrieval — built with desktop-native ambition.
Role
Solo — architecture, backend, retrieval, and interface.
Proof
- Persistent memory across sessions with local-first privacy.
- Document-grounded retrieval and streaming chat.
Stack
FastAPIReactTypeScriptLocal LLMsRAG
Field note · classified