Alumni Mentorship Android App with Interest Based Matching

MentorLink is a native Android app that connects college students with alumni mentors through interest-based matching. A student creates a profile, selects interest tags such as Android Development, Machine Learning or Startups, and the app ranks alumni mentors by Jaccard similarity — shared interests divided by all distinct interests — so every recommendation carries an explainable score. Each mentor profile shows the exact overlap arithmetic, the shared tags, and weekly availability; the student then sends a mentorship request with a topic and message and books a session slot. The backend is Firebase Authentication plus Cloud Firestore, with collections for students, mentors, requests and sessions. The problem it solves is concrete: most colleges run alumni mentorship through spreadsheets, WhatsApp groups or one annual meet, so a student who needs guidance on placements or interviews rarely finds the right senior. MentorLink replaces that ad-hoc process with a searchable, ranked directory and a structured request-to-session flow.

From the student's first tap to a booked session, MentorLink runs a closed loop: sign-up, tag-based profiling, ranked discovery, request, and booking. The student signs in with Firebase Authentication, then completes onboarding by entering name, branch and year and selecting interest tags; the profile is saved as a document in the students collection of Cloud Firestore. Alumni profiles — name, role, company, graduation batch and expertise tags — live in the mentors collection, seeded by the college. When the directory opens, the app computes the Jaccard similarity between the student's tag set and every mentor's tag set, sorts mentors by score, rendering each card with its percentage and shared-interest count. Opening a profile reveals the arithmetic alongside the mentor's availability for the next seven days. The student picks a topic chip, writes a message, optionally selects a slot, and sends the request; the mentor's accept or decline updates the request status, and an acceptance creates a booked session under Requests.

1. The student registers with email and password through Firebase Authentication and completes the onboarding form: name, branch, year and interest tags.
2. The profile, including the tag set, is written as a document in the students collection in Cloud Firestore.
3. Alumni profiles with expertise tags and weekly availability are stored in the mentors collection (seeded by the college in the shipped build).
4. Opening the directory triggers the matching engine: Jaccard similarity is computed between the student's tags and each mentor's tags, and mentors are sorted by score.
5. Each mentor card displays the match percentage, shared-interest count and a progress bar; search filters by name, role, company or skill.
6. The mentor detail screen shows the exact match arithmetic (shared ÷ distinct = score), shared tags highlighted, bio, and day/slot pickers for the next 7 days.
7. The student sends a request with a topic chip, message and optional slot; it is stored in requests with status pending.
8. The mentor accepts or declines; an acceptance creates a sessions document, and the student sees the booked session with date and slot under Requests.

Tech stack:

• Kotlin + Jetpack Compose
• Firebase Authentication
• Cloud Firestore
• Firebase Cloud Messaging (optional)
• Android Studio
• Material Design 3
• JUnit (unit tests)
• Git

The app stands on four pillars: an explainable matching engine, transparent mentor profiles, a structured request flow, and two-tap session booking. The matching engine computes Jaccard similarity live on the device — shared tags divided by distinct tags — so a student sees exactly why one mentor ranks at 50% and another at 43%, with the shared interests named on every card. Mentor profiles go beyond a name list: role, company, graduation batch, location, expertise tags with shared ones highlighted, a short bio, and a mentoring record. The request flow keeps outreach professional: topic chips such as career guidance, resume review and interview preparation, a personal message field, and an optional preferred slot travel together in one request carrying pending, accepted or declined status. Finally, each mentor publishes availability for the next seven days, so booking a session means picking a day and a slot, with every confirmation tracked on a single Requests screen.

• [Interest-tag onboarding] The student signs up and picks interests from a 16-tag vocabulary (Android Development, Machine Learning, Cloud Computing, Startups, Placements and more). The tag set drives all matching and can be edited later from the profile screen.
• [Jaccard interest matching] Every mentor is scored as shared tags ÷ distinct tags across both sets, computed live in the app and rendered as a percentage with a progress bar. Measured on the demo build, a student tagged Android Development, Machine Learning, Cloud Computing, Startups and Placements scores Priya Nair 0.50 (3 shared of 6 distinct interests), Divya Menon 0.43 (3 of 7), three alumni 0.29, two 0.125, and a zero-overlap profile 0.00 — the ranking is arithmetic, not a black box.
• [Ranked alumni directory with search] Mentor cards show photo-initial avatar, name, role, company, shared-interest count, rating and mentees mentored, sorted by match score. A search bar filters by name, role, company or skill.
• [Transparent match breakdown] Opening a profile reveals the full calculation — for example J = 3/6 = 0.50 — with shared tags highlighted in green, so the student trusts why this mentor was recommended.
• [Mentorship request flow] Topic chips (career guidance, resume review, interview preparation, higher studies, startup advice, technical deep-dive) plus a personal message and an optional preferred slot go out as one request; statuses move from pending to accepted or declined.
• [Session slot booking] Each mentor's next seven days of availability render as day and slot pickers. Accepting a request creates a booked session entry with date, slot and topic under Requests.
• [Firebase backend] Firebase Authentication (email and password sign-in) plus Cloud Firestore collections for students, mentors, requests and sessions, with security rules so users only read and write their own data. The interactive demo preview simulates this backend with on-device storage.
• [Optional: push reminders] Firebase Cloud Messaging reminders for upcoming sessions can be added as an optional extension; the base build tracks sessions without them.

• Complete Kotlin + Jetpack Compose Android source code
• Firebase setup guide: Authentication, Firestore collections and security rules
• Seeded demo data: 8 alumni profiles, 16-tag vocabulary, availability generator
• Matching-engine module (Jaccard) with unit-test notes
• Interactive HTML demo preview of all app screens
• Project report PDF (background, architecture, matching algorithm, methodology, results)
• PPT presentation for final review
• Viva Q&A preparation document (Kotlin, Compose, Firebase, Jaccard similarity, app architecture)