ANVIKA

Context & Problem

Most women's health apps treat the body like a clock: fixed-cycle calendars, generic ovulation windows, and a business model built on selling intimate data to third-party advertisers. The result is an industry-wide trust deficit. Women with irregular cycles, PCOS, or perimenopause are systematically underserved by apps that assume a textbook 28-day cycle exists for everyone. Beyond the biological inaccuracies, there is a deeper structural problem — sensitive physiological and emotional data is routinely uploaded to centralized servers, shared with data brokers, and stored indefinitely without meaningful user consent or control. ANVIKA was built to challenge both of these failures simultaneously: bring genuine predictive intelligence to cycle tracking, and do it without ever compromising the user's privacy.

Solution & Approach

ANVIKA is architected as a local-first, offline-capable Flutter application. All user data — cycle logs, mood entries, symptom records, and habit streaks — is stored exclusively on-device using Drift, a reactive SQLite wrapper for Dart that provides type-safe database access and migration support. No data is ever uploaded to a remote server, which means the app continues working fully without an internet connection, and there is zero risk of a server-side data breach. The cycle prediction engine is built using adaptive statistical modelling rather than a fixed calendar assumption. It learns each user's individual baseline from logged data, adjusting predictions as patterns shift over time. Mood and symptom entries feed into a lightweight correlation layer that identifies recurring connections between hormonal cycle phases and emotional or physical states — giving users genuinely personalised insights rather than generic tips. Firebase is used only for authentication and crash reporting metadata, with no personally identifiable health data ever leaving the device.

Key Features & Highlights

• Local-first encrypted data storage using Drift (SQLite): all cycle, mood, and symptom data is persisted entirely on-device with reactive streams, type-safe queries, and zero remote exposure.
• Adaptive cycle prediction engine that builds a personal baseline from the user's own logged history, handling irregular cycles, PCOS patterns, and perimenopause shifts that fixed-calendar apps cannot model.
• Mood and symptom correlation layer that identifies statistically recurring links between cycle phases and emotional or physical states, surfacing personalised insights on the insights dashboard.
• Micro-habit retention system with streak tracking, contextual nudges, and phase-aware recommendations — suggesting habits timed to cycle energy levels rather than arbitrary daily reminders.

Results & Impact

ANVIKA is live on the Google Play Store. The local-first, zero-telemetry architecture means the app genuinely earns user trust as a differentiator — not just as a marketing claim. Early feedback highlights the offline reliability and the absence of data-sharing prompts as key reasons users prefer it over established competitors. The project also sharpened production skills in Drift schema migrations, reactive UI patterns in Flutter, and designing predictive systems that degrade gracefully when data is sparse.