FinLedger — Android Personal Finance & Wealth Management Engine
Local-first, zero-knowledge Android personal finance engine with SQLCipher encryption, on-device bank SMS detection, and zero double-counting accounting.
Technologies & Architecture Stack
Project Overview
FinLedger is a local-first, zero-knowledge personal finance management system engineered for Android (Kotlin 2.0, Jetpack Compose, Room v18, and SQLCipher). Unlike mainstream commercial finance apps that compromise financial privacy by scraping credentials or uploading raw banking data to remote servers, FinLedger executes 100% of ledger calculations, transaction classification, and SMS parsing on-device. The platform solves the fundamental accounting flaw present in popular budgeting apps: double-counting credit card liabilities and bill settlements. Built with Clean Architecture, an immutable domain LedgerEngine, 256-bit AES SQLCipher encryption anchored to Android KeyStore hardware (TEE/StrongBox), and zero-knowledge Google Drive cloud sync (REST API v3), FinLedger bridges deterministic financial precision with fluid 60 FPS Material 3 design.
The Engineering Problem
Mainstream personal finance applications suffer from critical architectural and mathematical flaws: 1. Severe Privacy Leakage: Banking SMS notifications and transaction histories are continuously uploaded to third-party analytics servers. 2. Flawed Financial Mathematics: Most apps double-count credit card purchases and subsequent bill payments, distorting liquid spendable cash and true net worth. 3. High Data Entry Friction: Manual expense logging is tedious, while automated cloud scrapers frequently fail or break due to bank security changes. 4. Database & State Inconsistency: Concurrent balance mutations, app process kills during imports, and multi-device synchronizations often corrupt ledger state.
The Solution
Engineered an offline-first Android financial operating system rooted in mathematical determinism and hardware-backed security: 1. Deterministic Ledger Engine: Developed a pure Kotlin Posting rules engine that cleanly isolates liquid cash, credit card liabilities, and illiquid wealth with zero double-counting. 2. Local On-Device Regex SMS Ingestion: Built a background broadcast receiver and inbox scanner with an early OTP-rejection shield that identifies financial transactions from 10+ Indian/global banks and UPI apps without sending a single byte over the network. 3. Cryptographic Hardware Vault: Encrypted the SQLite database at rest using SQLCipher 256-bit AES-CBC, derived from keys protected within the Android KeyStore (TEE / StrongBox) and salted PBKDF2 PIN authentication (50,000 iterations). 4. Zero-Knowledge Google Drive AppData Sync: Designed a lightweight REST API v3 sync engine that uploads client-side encrypted snapshots to the user's hidden Google Drive appDataFolder with bidirectional conflict resolution. 5. Client-Side Document Generation: Integrated native PdfDocument statement rendering and universal RFC-4180 CSV import/export with pre-import safety snapshots and 1-tap rollback.
System Architecture & Data Flow
Built on Clean Architecture and Unidirectional Data Flow (UDF) across three decoupled layers: - Presentation Layer: 24 Jetpack Compose navigation routes themed with Material 3. ViewModels expose immutable StateFlow collected via collectAsStateWithLifecycle. Startup data fetching is decoupled to ensure cold-boot primary balance emission in <3ms. - Domain Layer: Pure Kotlin business logic with zero Android/Room dependencies. Includes LedgerEngine (deterministic replay and delta calculation), Posting.kt (canonical accounting rules), SmartCategoryClassifier (merchant heuristic learning), and SyncMergeEngine. - Data Layer: Room Database v18 (21 Entities, 18 DAOs) wrapped by SQLCipher AES-256. A unified FinanceRepository facade coordinates specialized repositories (Transaction, Account, Category, Budget, Lending, and Sync). WorkManager handles periodic background budget threshold notifications, due-date alerts, and daily sync.
Core Engineering Features
Technical Challenges & Overcoming Them
Results & Impact
- 158 automated unit tests across 21 test suites executing with 100% pass rate. - Cold-boot primary balance emission in <3ms on mid-range Android hardware. - 0% double-counting across multi-account transfers, credit settlements, and expense logging. - 256-bit AES database encryption at rest combined with 50,000 PBKDF2 iterations for zero plaintext leak. - 100% offline, zero-telemetry architecture with zero raw transaction data leaving the device. - Release candidate (v2.0.2, Build 8) verified with R8 minification, ProGuard log stripping, and WCAG AA contrast compliance.
Key Takeaways
- Mastered Clean Architecture and Unidirectional Data Flow (UDF) in Jetpack Compose, learning how to isolate pure business logic from UI and persistence frameworks. - Deepened expertise in cryptographic key management, TEE hardware security, side-channel attack mitigation (constant-time verification), and SQLCipher encryption. - Gained advanced experience in asynchronous Android systems: broadcast receivers, WorkManager scheduled tasks, and high-performance Kotlin Coroutines/Flow state modeling. - Learned the nuances of offline-first synchronization protocols, monotonic versioning, and deterministic state machines for conflict resolution without a central backend server.
Future Roadmap
- On-Device Receipt & Invoice OCR: Integrate Google ML Kit Text Recognition for automated receipt itemization and camera-to-transaction parsing. - Mutual Fund NAV Sync Engine: Daily automated background portfolio valuation via official AMFI India public APIs. - Advanced Financial Modeling: Multi-scenario retirement and FIRE (Financial Independence, Retire Early) forecasting with monte carlo simulations. - Excel (.xlsx) Multi-Tab Export: Native streaming generation of formatted multi-account workbooks.