# LifeOS — One Account. One Search. One AI.

> Category: Full-Stack · AI | Period: Planned — after CineRoulette | Metric: 20+ unified modules
> Canonical URL: https://www.arman-portfolio.online/projects/lifeos/
> Live Demo: https://lifeos-demo.vercel.app/
> Source Repository: https://github.com/arman080325/lifeos.git

## Overview
A privacy-first personal operating system unifying 20+ life modules — tasks, notes, calendar, health, finance, career, and more — behind a single Ctrl+K universal search and a context-aware AI layer, instead of a dozen isolated apps. Queued to start right after CineRoulette ships publicly.

## Architecture & Tech Stack
Next.js, NestJS, PostgreSQL, Prisma, Redis, Vector DB, WebSockets

## Case Study

### The Problem
Notes, tasks, health, finance, and career tools each solve one isolated problem in their own silo — switching between a dozen single-purpose apps means the same person's data never actually connects, and nothing proactively surfaces what matters.

### Architecture & Approach
LifeOS is scoped around one philosophy — one account, one search, one AI, everything connected through a shared data model instead of isolated apps — spanning 20+ modules (Dashboard, Tasks, Notes, Calendar, Goals, Habits, Learning, Coding, Career, Finance, Health, a zero-knowledge Password Vault, and more) behind a single hybrid lexical-and-semantic Ctrl+K search, with an AI layer doing RAG, embeddings, and long-term memory across every module.

### Engineering Challenges & Solutions
- Designing a shared data model expressive enough for wildly different modules — a workout set, a password credential, a calendar event — without collapsing into a meaningless generic 'item' table
- Keeping the zero-knowledge password vault's encryption model cleanly separated from the AI layer's semantic indexing, since the two modules have opposite requirements around what the server is allowed to read
- Scoping a genuinely large 20+ module vision into a realistic MVP → Productivity Suite → Career Hub → Documents & Finance → Health → Family → Plugins roadmap instead of trying to build all of it at once

### Outcome & Results
Design-stage — the full module list, database design, AI architecture, and an MVP-first roadmap are locked in a living v1 document, deliberately queued to start only once CineRoulette is deployed publicly, so it gets full, undivided build attention.

### Lessons Learned
Not yet started — the plan is the current artifact. Keeping the roadmap honest about a 200+ page 'complete handbook' as the eventual full vision, while treating the MVP as the only thing that matters for the first commit.

---
Author: Arman Ahemad Khan (https://www.arman-portfolio.online/)
Role: Application Security & Full-Stack Developer
Education: Final-year B.Tech CSE at Silicon University, Bhubaneswar (CGPA 9.15, graduating May 2027)