Professional Project
Deepstory
Overview
Deepstory is a social media platform built to let users share and discover rich, story-driven content. As the sole developer, I owned the full lifecycle — from initial architecture decisions to production deployment — integrating custom feed algorithms, robust media handling, and comprehensive user analytics.
Role
Full Stack Android Developer at Designare Solutions
Problem
The startup needed a production-ready social media Android application built from scratch. Beyond basic CRUD operations, the platform required a custom content recommendation engine, reliable background video uploads, and deep user behavioral tracking to guide business decisions—all delivered by a single engineer.
Solution
I designed a layered MVVM Android architecture using Kotlin and RxJava for reactive event handling (such as video upload progress). The backend was powered by a Node.js REST API and PostgreSQL, deployed on GCP. I implemented cron jobs to calculate per-video engagement metrics and drive a proprietary, weighted-scoring recommendation algorithm, while integrating Firebase and Amplitude for secure authentication, crash reporting, and behavioral analytics.
Architecture
A native Android client (Kotlin/MVVM) communicating with a Node.js REST API, backed by PostgreSQL for relational data and GCP Cloud Storage for media. Firebase and Amplitude handle auth and analytics, while RxJava manages asynchronous client events.
Key Design Decisions
- MVVM architecture on Android using Kotlin, LiveData, and ViewModel for reactive UI state
- RxJava implemented for publisher events and asynchronous tracking, specifically managing background video upload progress
- RESTful Node.js API handling core social features (likes, comments, video uploads) backed by PostgreSQL
- Maintained a centralized Postman workspace documenting the entire REST API ecosystem to ensure seamless integration and testing
- Custom cron jobs implemented to calculate per-video engagement counts (views, likes, comments)
- Proprietary, category-based recommendation engine utilizing a weighted scoring system to tailor user-specific feeds
- GCP Cloud Storage integration for scalable user-uploaded media management
- Firebase suite integrated for Authentication, Crashlytics, and core user tracking
- Amplitude integrated for deep behavioral analytics to capture and analyze user engagement patterns
Challenges
- Designing a scalable social graph schema in PostgreSQL without over-engineering for a startup's early user base
- Engineering a proprietary, weighted recommendation algorithm and orchestrating cron jobs to process per-video metrics efficiently
- Handling heavy media upload flows on Android using RxJava and background workers while keeping the UX responsive
- Sole ownership across Android, API, DevOps, and analytics integration requiring intense context-switching
Impact
- Delivered a fully functional, data-driven social media Android app from zero to production as a single engineer
- Enabled tailored user experiences and targeted content delivery through the custom weighted-scoring recommendation engine
- Provided leadership with actionable user behavior insights via Amplitude and Firebase integration
- Established a reliable GCP-based deployment pipeline reused across subsequent projects