Portfolio project by beingAnujChaudhary · Phase 2: Web MVP

Read smarter.
Think deeper.
Stay offline.

aetherRead is a production-grade PDF reader with a local AI brain, distraction-free reading themes, and real-time cross-platform sync — built for students, researchers, and academics who demand privacy.

Try It FreeView Project Page
PDF Reader
Local AI Brain
Offline First
Privacy First
100%
Offline Capable
6
Reading Themes
RAG
Local AI Engine
0
Data Sent to Cloud
Scroll
Everything you need

Built for serious readers

Every feature is designed around one principle: your reading flow should never be interrupted — by internet, by apps, or by distractions.

🌐 Web App🤖 Android🖥️ Windows Native
Core

True Offline-First

Every feature works without internet. PDFs, annotations, and reading progress stored locally via IndexedDB using Dexie.js.

AI

Local AI Brain

Chat with your PDFs using Ollama-powered LLMs. ChromaDB vector store. Your documents never leave your network.

UX

6 Comfort Themes

Dark Abyss, Book Paper, Sepia Sands, Focus Punch, Monochrome, Garden Sage — hand-crafted reading themes that switch instantly.

Notes

Page-Level Annotations

Attach rich Markdown notes to any page, categorized as Important, Definition, Question, Revision, or Quote.

Annotate

13 Annotation Tools

Smart Pen, Text Highlight, Freehand Draw, Text Underline, Strikethrough, Squiggly, Sticky Note, Callout, Eraser, Multi-select — with live color picker.

Android

Native Android App

Kotlin + Jetpack Compose. Modern UI: Files, Toolbox, My AetherRead, Aether Sign tabs. Annotation tools, comfort themes, reader toolbar.

Windows

Native Windows App

C# WPF (.NET 8) — zero Electron overhead. Google Pdfium rendering engine, left navigation rail, native file dialogs. Windows 10/11.

Tools

Toolbox Suite

Scan Document, Image to PDF, eSign PDF, OCR Text Recognition, Convert from/to PDF, and Manage PDF — all in one unified toolbox.

Sync

Decoupled Sync

LWW strategy for reading position. Two-way annotation merge. Soft-delete pattern ensures zero data loss across devices.

Analytics

Reading Analytics

WPM tracking, reading streaks, daily goals, and predictive "Time Left in Chapter" algorithm.

Security

Privacy First

Row Level Security on Supabase, TLS 1.3 in transit, zero frontend API key exposure, full JSON export. Your documents stay yours.

Study

Revision Packs

Auto-compile highlights into Markdown study guides, Anki CSV flashcards, or Obsidian vault exports.

Native Android App

Desktop-grade tools in your pocket.

We rebuilt the reading experience for Android from the ground up using Kotlin and Jetpack Compose. It brings fluid, memory-safe PDF rendering and powerful annotation tools completely offline.

Modern Annotation Pill

Floating bottom pill for quick access to drawing tools: Smart Pen, Highlight, Strikethrough, and Callout.

Toolbox Suite

A unified grid for document operations: Scan, Convert, Image to PDF, Merge, and Manage files natively.

Comfort Themes

6 carefully crafted color schemes applied directly to the PDF canvas, ensuring comfortable reading at any time.

Download App (ZIP)
✎
H
S
✕
Download

Available on All Platforms

Download aetherRead on your preferred platform. All versions share the same open-source codebase under the MIT license.

Android

Android 8.0+ (Oreo)
Source
  • Native Jetpack Compose UI
  • Offline-first PDF reader
  • Freehand, highlight & annotation tools
  • AI Brain integration
  • Dark mode & comfort themes
Download App (ZIP)

Windows

Windows 10 / 11 (Native WPF)
Source
  • Truly native WPF / C# app (.exe)
  • No browser or Electron overhead
  • Pdfium C++ rendering engine
  • Left-nav rail — Native UI
  • Native Windows file picker & dialogs
Download for Windows (ZIP)

Just want to try it?

Open the progressive web app instantly in your browser — no install required.

Web RepoOpen Web App
Comfort Engine

6 reading themes.
Zero eye strain.

Each theme is scientifically crafted for different lighting conditions and reading sessions. Switch instantly — your eyes will thank you.

research_paper.pdf — Page 4 of 42
🌑 Dark Abyss

Chapter 4: Attention Mechanisms

The transformer architecture introduced in "Attention is All You Need" (Vaswani et al., 2017) revolutionized natural language processing by replacing recurrent layers with self-attention mechanisms.

Self-attention computes:

Attention(Q, K, V) = softmax(QK^T / √d_k) × V

where Q, K, V are query, key, and value matrices derived from the input sequence.

📌 Important — Page 4
Engineering

Production-grade architecture

aetherRead uses enterprise-grade, decoupled architecture across three repositories with a strict offline-first, privacy-first engineering philosophy.

Web Frontend

  • Next.js 14 (App Router)
  • TypeScript
  • TailwindCSS
  • Dexie.js (IndexedDB)
  • Zustand
  • @react-pdf-viewer/core

Android App

  • Kotlin + Jetpack Compose
  • Room Database
  • Hilt DI
  • Retrofit
  • AndroidPdfViewer
  • WindowSizeClass

Backend API

  • FastAPI + Python 3.11
  • PyMuPDF / Marker-pdf
  • Pydantic v2
  • Docker
  • GitHub Actions CI/CD
  • Uvicorn

AI / ML Pipeline

  • Ollama (llama3.2:3b)
  • ChromaDB Vector Store
  • LangChain
  • RAGAS Evaluation
  • RAG with Citations
  • Local Mesh Networking

Database & Sync

  • Supabase (PostgreSQL)
  • pgvector extension
  • Supabase Realtime
  • Row Level Security
  • LWW Conflict Resolution
  • Soft-Delete Pattern

Architecture

  • Offline-First Design
  • Three-Repo Monorepo
  • ADR Documentation
  • Open-Core Model
  • PWA Support
  • WCAG AA Accessible

Three-repository structure: aetherRead-web · aetherRead-android · aetherRead-api — decoupled for independent deployment, versioning, and open-source release. Web and Android are MIT/AGPL-3.0. API is proprietary.

Development Roadmap

From MVP to production

Six focused phases, each delivering real value. Currently in Phase 5.

Phase 1✓ Complete

Foundations

  • Project architecture & ADRs
  • Repository setup (3 repos)
  • Supabase schema hardening
Phase 2✓ Complete

Web MVP

  • PDF rendering with offline storage
  • 6 reading themes (Comfort Engine)
  • Reading analytics & WPM tracking
  • Cold start onboarding
Phase 3✓ Complete

Native Android

  • Native PDF rendering with memory safety
  • Tablet layout (NavigationRail)
  • Offline-first with Room DB
Phase 4✓ Complete

Windows Native WPF

  • PdfiumViewer C++ rendering
  • Native WPF window & dark mode
  • Modern navigation rail
Phase 5⚡ Active

Decoupled Sync

  • LWW reading position sync
  • Two-way merge for annotations
  • Sync status indicator UI
Phase 6Upcoming

Local AI Brain

  • PDF extraction with Marker
  • Ollama integration
  • RAGAS evaluation pipeline
Try the App Now

Offline-first · No account required for local reading