Learn2Hire is an AI-powered Android learning and interview preparation application designed to bring structured study, assessment, adaptive practice, AI tutoring, interview preparation, planning, and progress tracking into one unified mobile experience.
Built using Kotlin, Jetpack Compose, Firebase Authentication, Cloud Firestore, Firebase AI Logic, and Gemini, Learn2Hire combines deterministic academic assessment with AI-assisted learning. Fixed quizzes remain the authoritative measure of topic performance, while AI Practice, the AI Tutor, and Interview Coach provide personalized training and feedback.
The application currently includes structured learning content for Machine Learning, Web Application Development, and Automata Theory & Compiler Design, with persistent user progress and activity stored securely per Firebase-authenticated account.
- 📚 Structured Study Modules: Learn through subjects, topics, lesson content, key concepts, estimated study time, and completion tracking.
- 📝 Fixed Quiz Assessments: Deterministic topic quizzes provide consistent and authoritative learning-performance scores.
- 🧠 Adaptive AI Practice: Generates topic-aware practice questions with difficulty adapted to the learner's quiz performance.
- ✨ Local Practice Fallback: Uses authored topic-specific question banks when Gemini is temporarily unavailable or rate-limited.
- 🤖 AI Tutor: Ask contextual questions about study topics and receive Gemini-powered explanations grounded in the learning material.
- 💬 Persistent Tutor History: AI Tutor conversations are stored in Firestore and remain available across sessions.
- 🎤 Interview Preparation: Practice role- and type-based interview questions with deterministic key-point evaluation.
- 💡 AI Interview Coach: Gemini provides qualitative feedback, improvement suggestions, and stronger example answers.
- 📅 Study Planner: Create tasks, track completion, and record study sessions.
- 📊 Progress Analytics: View study completion, quiz performance, AI Practice activity, interview performance, and topic-level learning signals.
- 🎯 Personalized Study Insights: Home recommendations adapt to the learner's recorded study and assessment activity.
- 🔐 Firebase Authentication: Supports secure account-based authentication, including Google Sign-In.
- ☁️ Cloud Persistence: Study progress, quizzes, AI sessions, planner data, profile information, and chat history persist through Firestore.
- 🌗 Theme Support: Light, Dark, and System theme modes with a custom Learn2Hire visual identity.
Learn2Hire currently contains 3 subjects and 10 structured topics:
| Subject | Topics |
|---|---|
| Machine Learning | Introduction to ML, Classification, Regression, Naïve Bayes |
| Web Application Development | Client and Server, Node.js and Express, SQL and NoSQL |
| Automata Theory & Compiler Design | Finite Automata, DFA and NFA, Regular Expressions |
Each topic contains lesson material, key concepts, estimated study duration, progress tracking, fixed assessment, AI Practice access, and contextual AI Tutor access.
Learn2Hire deliberately separates assessment from AI-assisted practice.
| Component | Purpose | Source | Used for Authoritative Learning Level? |
|---|---|---|---|
| Fixed Quiz | Consistent topic assessment | Authored question bank | ✅ Yes |
| AI Practice | Adaptive practice and improvement | Gemini + local fallback | ❌ No |
| AI Tutor | Explanation and study assistance | Gemini + study-material fallback | ❌ No |
| Interview Coach | Qualitative interview improvement | Gemini | ❌ No |
This prevents generative AI variability from directly determining the learner's official topic-performance score.
Quiz performance is used to classify the learner's topic state:
| Quiz Performance | Learning State | AI Practice Level |
|---|---|---|
| No assessment | Not Assessed | Standard |
| Below 60% | Needs Practice | Foundation |
| 60–79% | Developing | Intermediate |
| 80% and above | Strong | Challenge |
AI Practice remains a secondary improvement signal and is displayed separately from fixed quiz assessment in Progress.
Study Topic
│
┌─────────────┼─────────────┐
↓ ↓ ↓
Fixed Quiz AI Practice AI Tutor
│ │ │
↓ ↓ ↓
Deterministic Gemini AI Gemini AI
Assessment │ │
│ ┌─────┴─────┐ ┌─────┴─────┐
│ ↓ ↓ ↓ ↓
│ Generated Local Study AI Response
│ Questions Fallback / Fallback
│
↓
Authoritative Score
│
└─────────────┐
↓
Learning Profile
│
↓
Adaptive Recommendations
Firebase AI Logic is used to access Gemini without embedding a raw Gemini API key directly in the Android source code.
Select Role & Interview Type
↓
Interview Questions
↓
User Response
↓
Deterministic Key-Point Check
↓
Coverage Score
↓
Gemini AI Coach
↓
What Went Well / Improvements
/ Improved Answer
↓
Firestore Interview Session
The deterministic key-point evaluation remains the measurable interview score, while Gemini provides qualitative coaching.
Each authenticated user owns their learning data under their Firebase UID.
users/{uid}
│
├── name
├── email
├── photoUrl
├── course
├── semester
├── authProvider
├── createdAt
│
├── studyProgress/{topicId}
├── quizAttempts/{attemptId}
├── aiPracticeSessions/{sessionId}
├── tutorChats/{chatId}
├── interviewSessions/{sessionId}
├── plannerTasks/{taskId}
└── plannerSessions/{sessionId}
Firestore security rules restrict user data access to the authenticated owner of that user document.
Authentication
↓
Home
│
├───────────────┬─────────────────┐
↓ ↓ ↓
Study Interview Planner
│ Prep │
↓ ↓
Subject → Topic Tasks &
│ Study Sessions
├──────────────┐
↓ ↓
Fixed Quiz AI Features
│ ┌────┴────┐
↓ ↓ ↓
Assessment AI Practice AI Tutor
│ │ │
└──────────┼─────────┘
↓
Progress
│
↓
Personalized Insight
| Category | Technology |
|---|---|
| Platform | Android |
| Programming Language | Kotlin |
| UI | Jetpack Compose, Material 3 |
| Architecture | ViewModel + Repository + Reactive UI State |
| Navigation | Jetpack Navigation Compose |
| Authentication | Firebase Authentication |
| Google Sign-In | Credential Manager + Google Identity |
| Database | Cloud Firestore |
| Generative AI | Firebase AI Logic + Gemini |
| Async Operations | Kotlin Coroutines |
| Local Preferences | SharedPreferences |
| Build System | Gradle Kotlin DSL |
| Minimum Android Version | API 23 |
| Target SDK | API 37 |
Learn2Hire/
│
├── app/
│ ├── build.gradle.kts
│ ├── google-services.json
│ │
│ └── src/main/
│ ├── AndroidManifest.xml
│ ├── java/com/learn2hire/app/
│ │ │
│ │ ├── data/
│ │ │ ├── local/
│ │ │ │ ├── StudyContent.kt
│ │ │ │ ├── SampleQuizData.kt
│ │ │ │ └── SampleInterviewData.kt
│ │ │ │
│ │ │ ├── model/
│ │ │ └── repository/
│ │ │
│ │ ├── ui/
│ │ │ ├── auth/
│ │ │ ├── components/
│ │ │ ├── home/
│ │ │ ├── interview/
│ │ │ ├── navigation/
│ │ │ ├── planner/
│ │ │ ├── practice/
│ │ │ ├── profile/
│ │ │ ├── progress/
│ │ │ ├── quiz/
│ │ │ ├── study/
│ │ │ ├── theme/
│ │ │ └── tutor/
│ │ │
│ │ └── MainActivity.kt
│ │
│ └── res/
│
├── gradle/
├── build.gradle.kts
├── settings.gradle.kts
└── README.md
The interface gallery below uses the screenshots included in the repository's screenshots/ directory.
| Login | Register |
|---|---|
 |
 |
| Home | Home — Continued |
|---|---|
 |
 |
| Study | Subject Detail |
|---|---|
 |
 |
| Fixed Quiz | AI Practice |
|---|---|
 |
 |
| Interview Setup | Interview Practice |
|---|---|
 |
 |
| Progress Overview | Topic Performance |
|---|---|
 |
 |
Before running the project, install:
- Android Studio
- JDK 11 or compatible JDK configured for the project
- Android SDK required by the project
- A Firebase project configured for the Android package
com.learn2hire.app - Firebase Authentication and Cloud Firestore
- Firebase AI Logic with the Gemini Developer API backend configured
Clone the repository:
git clone https://github.com/abhijithshetty12/Learn2Hire.gitNavigate into the project:
cd Learn2HireOpen the project in Android Studio and allow Gradle synchronization to complete.
Add your Firebase Android configuration file at:
app/google-services.json
Do not add release keystores, keystore passwords, or other private signing credentials to the repository.
On Windows PowerShell:
.\gradlew assembleDebugThe debug APK will be generated under:
app/build/outputs/apk/debug/
Run Android lint:
.\gradlew lintDebugFor a release build, configure your own Android signing key and generate a signed APK or Android App Bundle through Android Studio.
Pre-built releases can be published through the repository's Releases section:
For security, the release signing keystore is not included in the repository.
Learn2Hire follows several important security practices:
- Firebase user data is isolated by authenticated user ID through Firestore security rules.
- The Gemini API credential is not hard-coded directly into the Android source.
- AI access is implemented through Firebase AI Logic.
- Android release signing keys and passwords must remain outside version control.
- Local machine configuration such as
local.propertiesshould not be committed. - Firebase Authentication is used to identify users before accessing personal learning data.
The project has been validated through:
Gradle Clean Build ✓
Debug APK Compilation ✓
Android Lint ✓
Runtime Navigation ✓
Firebase Persistence ✓
Authentication ✓
Study Progress ✓
Fixed Quiz Assessment ✓
AI Practice ✓
AI Tutor & Chat History ✓
Interview Preparation ✓
Planner ✓
Profile & Themes ✓
Sign-out / Sign-in Data ✓
The main objective of Learn2Hire is to create a unified learning and career-preparation platform that combines structured academic study with responsible AI assistance.
Instead of allowing generative AI to determine academic performance directly, Learn2Hire maintains deterministic fixed assessments for measurable learning progress while using AI for practice generation, tutoring, explanations, and interview coaching.
The system demonstrates how traditional application logic, cloud persistence, learning analytics, and generative AI can work together in a practical Android application.
- The current structured learning library contains three subjects and ten topics.
- Fixed quiz coverage is limited to the learning content currently bundled with the application.
- Gemini responses depend on network connectivity, service availability, and applicable API quotas.
- AI-generated content can vary between requests and should be treated as learning assistance rather than an authoritative academic source.
- Local fallbacks provide continuity when AI generation is unavailable, but they are limited to the authored content bundled with the application.
- Interview evaluation uses predefined key points for deterministic scoring and does not attempt to fully measure real-world interview performance.
- The current application is designed as an academic/project implementation rather than a production-scale learning management system.
- Expand the study library with additional engineering and placement subjects.
- Add richer personalized learning paths based on long-term performance.
- Introduce spaced repetition and revision scheduling.
- Add more extensive question banks and assessment modes.
- Support coding interview practice and programming challenges.
- Add resume analysis and role-specific preparation plans.
- Introduce richer progress visualizations and learning streaks.
- Add instructor/admin capabilities for managing learning content.
- Support notifications for planner tasks and revision reminders.
- Add offline-first caching for selected learning content.
- Expand AI grounding and retrieval over larger course materials.
- Publish through Google Play with production-ready monitoring and release infrastructure.
Abhijith Shetty
*AI & Machine Learning Student | Android & AI Developer *
"Building intelligent applications that turn learning into measurable progress and career preparation."
If you find Learn2Hire interesting or useful, consider giving the project a ⭐ on GitHub.
It helps support the project and encourages further development.