Android-first technology intelligence prototype for following meaningful developments instead of isolated headlines.
ShiftInView explores how a technology-news product can combine development clustering, evidence context, personalized discovery and community discussion without allowing popularity signals to redefine evidence quality.
Project status: portfolio / academic prototype. This repository is not presented as a production deployment or Play Store release. The bundled preview data is fictional and intended only for product demonstration.
Eight selected screens show the core product flow without attempting to catalogue every prototype surface.
| Personalization | Today briefing |
|---|---|
![]() |
![]() |
| Product principles | Development detail |
|---|---|
![]() |
![]() |
| Explore | Community |
|---|---|
![]() |
![]() |
| Reader profile | Research tools |
|---|---|
![]() |
![]() |
Technology coverage is fast, repetitive and fragmented across sources. A user may see many articles about the same underlying event while still missing what actually changed, why it matters and which sources add independent information.
ShiftInView is designed around developments, not individual headlines. Its core product question is:
How can a reader follow what materially changed while keeping social popularity separate from evidence quality?
- Development-first feed — related coverage can be grouped around an underlying development rather than shown as disconnected headlines.
- Evidence-aware context — source role, source independence and material-change signals are modeled separately from reader engagement.
- Personalized discovery — interests, follows and reading context can influence ranking without rewriting shared evidence state.
- Community layer — readers can discuss, follow profiles, save collections and react to posts, while those social signals remain separate from the intelligence score.
- Research-oriented tools — saved searches, watchlists, claim/evidence mapping, research rooms and forecast tracking explore deeper workflows beyond a standard feed.
- English / Turkish interface support — the product shell supports both languages. Some preview or source-originated content may remain in its original language; source translation is an explicit reading action rather than a silent rewrite.
The strongest portfolio flow is:
- Personalization — choose interface language and initial interests.
- Today — personalized technology briefing and development cards.
- Development detail — inspect what happened, why it matters, evidence context and related developments.
- Explore — search and filter by topic, entity, source and signal characteristics.
- Community / Profile — follow readers, discuss developments, save collections and inspect reader activity.
- Research tools — use watchlists, claim/evidence maps or forecast tracking for deeper analysis.
ShiftInView was developed as an AI-assisted, iterative portfolio prototype. I defined and refined the product concept, scope, feature priorities, UX requirements, acceptance criteria and final integration decisions. AI coding tools were used extensively to accelerate implementation, repetitive coding, refactoring experiments, test scaffolding and iteration across a product scope that would otherwise require a substantially longer manual development cycle.
I repeatedly reviewed the integrated application, identified regressions, revised or removed features that did not support the product direction, and used manual QA plus automated backend tests to validate behavior. This repository does not claim that every line of code was written manually; it is intended to demonstrate product/system thinking, technical experimentation, critical review and responsible use of modern development tools.
- product concept, requirements and feature prioritization;
- UX review and interaction decisions;
- system-behavior boundaries, including the separation of popularity from evidence quality;
- iterative QA, regression identification and acceptance/rejection of changes;
- final integration decisions, with extensive AI-assisted implementation and test iteration.
The rapid, broad prototyping approach also created architectural trade-offs. Those are documented rather than hidden; see docs/KNOWN_LIMITATIONS.md.
- Kotlin
- Android SDK 35 / min SDK 24
- AndroidX + Material Components
- RecyclerView / Navigation / Lifecycle components
- Firebase Auth, Firestore, Functions, Messaging and Storage
- ML Kit language identification and on-device translation
- Node.js 22
- Firebase Functions / Admin SDK
- Firestore security rules and indexes
- Backend policy and intelligence tests using Node's built-in test runner
Android UI
│
├── Feed / Explore / Community / Profile / Research surfaces
│
├── repositories + data-source boundaries
│
└── Firebase access layer
│
├── Authentication
├── Firestore
├── Cloud Functions
├── Messaging
└── Storage
Backend functions
│
├── ingestion / normalization
├── discovery and policy logic
├── evidence / signal processing
└── social and notification rules
More detail: docs/ARCHITECTURE.md.
A central design decision is that popularity is not evidence quality.
Reader ratings, comments, follower counts and post engagement can provide social context, but they are not allowed to directly increase source independence, evidence quality or the versioned ShiftInView Signal Score. This separation is intended to reduce the risk of a popular claim appearing more credible simply because it receives more engagement.
See docs/TECHNICAL_DECISIONS.md for the reasoning behind this and other implementation choices.
The public repository does not contain private production credentials. With the Firebase configuration values left blank, the application is designed to expose a fictional preview experience so the product can be inspected without a live private backend.
All bundled preview names, organizations, stories and metrics should be treated as synthetic demonstration data.
- JDK 17
- Android Studio with Android SDK 35
- Internet access for the first Gradle/dependency sync if dependencies are not already cached
- Clone or download the repository.
- Open the repository root in Android Studio.
- Allow Gradle sync to complete.
- Run the
appconfiguration on an emulator or Android device.
Windows command-line build:
.\gradlew.bat clean assembleDebugmacOS / Linux:
./gradlew clean assembleDebugThe debug APK is generated under app/build/outputs/apk/debug/ after a successful build.
The public Android configuration boundary is:
app/src/main/res/values/shiftinview_firebase_config.xml
The repository intentionally does not include service-account files, private keys, google-services.json, keystores or environment secrets. Production deployment would require a separately configured Firebase project and deployment of the relevant Functions and security rules.
The repository includes Android unit/instrumentation tests and backend tests. In the portfolio export used to prepare this repository, the backend suite completed with 98/98 tests passing.
Run backend tests from backend/:
npm testThe backend tests cover areas such as clustering/signal policy, source independence, discovery, saved searches, social policy, notification behavior, mention privacy and storage path ownership.
A GitHub Actions workflow at .github/workflows/backend-ci.yml runs JavaScript syntax checks and the backend policy/unit test suite on pushes and pull requests. It is intentionally scoped to the dependency-free policy tests and is not presented as a Firebase deployment or full Android build check.
app/ Android application
backend/ Firebase Functions and backend tests
rules/ Firestore indexes and security rules
admin-console/ Lightweight administrative tooling
screenshots/ Selected portfolio screenshots
docs/ Architecture, technical decisions, privacy and limitations
This is intentionally presented as a portfolio prototype rather than a production service. Important limitations include:
- no Play Store release or production SLA;
- fictional / synthetic preview content;
- production Firebase credentials and deployment are not included;
- English/Turkish interface support is substantial but not claimed as a perfect translation audit of every runtime-generated or preview string;
- the Android UI grew through iterative prototyping and would benefit from further consolidation into a smaller reusable component/navigation architecture before production scaling.
Full notes: docs/KNOWN_LIMITATIONS.md.
The prototype separates private reader state from shared intelligence state and includes account/privacy concepts such as blocking, export/delete flows and user-scoped saved content. See docs/PRIVACY_AND_DATA_USE.md.
This is a portfolio review release, not an open-source release. Source code, product design, documentation and visual assets are shared for academic and professional evaluation under an all-rights-reserved portfolio license. See LICENSE.md.







