CampusMarket is the GitHub repository for UniKart, a campus-first commerce platform built for SRM KTR students to buy, sell, rent, compare, message, negotiate, schedule meetups, and transact with more structure than informal resale groups.
CampusMarketis the repo name.UniKartis the product name used across the codebase and UI.
UniKart is not just a listing board. It is a full-stack campus marketplace system with role-aware dashboards, listing composition, in-app messaging, offer and hold flows, meetup scheduling, campus route planning, AI-assisted listing quality and moderation, notification lanes, admin tooling, PDF exports, and a database layer hardened with Supabase migrations, indexes, and row-level security.
If the goal is to present this as a major engineering project, the strongest angle is this: the repo shows product thinking, front-end system design, API design, data modeling, operational tooling, safety-aware UX, and third-party service integration in one coherent system.
- Built as a campus-first resale platform for SRM KTR students
- Covers buyer, seller, and admin/operator workflows
- Includes
18frontend page routes,14backend route groups, and17Supabase migrations - Implements listing creation, compare queues, favorites, saved filters, notification inboxes, messaging, offers, holds, meetups, moderation, analytics, and exports
- Uses graceful fallback behavior for AI and campus routing when provider credentials are unavailable
- Structured as a real multi-surface product, not a single-demo page or isolated CRUD assignment
Screens below are captured from the actual application UI.
These screenshots show that the project is broader than a marketplace homepage:
- browse, shortlist, listing-detail, and follow-up flows are all implemented
- messaging, safety, and support have dedicated product surfaces
- meetup planning is treated as a first-class workflow with route context
- notifications and account-state management are visible parts of the product system
- It solves a specific environment problem: campus resale is high-frequency, trust-sensitive, and often handled badly through fragmented messaging apps.
- It treats safety as a product feature, not a disclaimer. Chat moderation, listing moderation, reporting, policy surfaces, blocked-user logic, and moderation workflows are built into the system.
- It goes beyond simple listing CRUD. The repo models negotiation, holds, schedules, notifications, recent activity, saved filters, compare queues, and operational analytics.
- It includes admin and operator surfaces, which makes the product feel like a platform rather than a student-only storefront.
- It demonstrates engineering maturity through typed contracts, validation, migrations, role checks, database policies, index strategy, fallbacks, caching, and export flows.
| Area | What is implemented |
|---|---|
| Marketplace discovery | Listing browse, detail pages, compare queue, favorites, recently viewed, trust-ranked listing views, saved search alerts |
| Seller experience | Guided sell studio, draft saves, publish flow, inventory count, rent and bundle modes, media previews, listing quality checks |
| Messaging and negotiation | Listing-linked chat, contact redaction, risky-message screening, offer and counter-offer flows, soft holds, thread termination |
| Meetup coordination | Proposed meetup creation, two-sided confirmation state, route persistence, location search, campus walking route previews |
| Trust and safety | Listing reports, moderation actions, policy surfaces, support flows, audit-friendly moderation history, safety notifications |
| Admin tooling | Moderation dashboard, platform analytics, AI diagnostics, provider event visibility, evidence packet PDF export |
| AI layer | Listing assist, search assist, listing moderation, chat moderation, persisted AI artifacts, provider-event telemetry |
| Export and documents | Listing packet PDF export, moderation evidence packet export, frontend PDF helpers for richer user-facing artifacts |
| Platform signals | Notification inboxes, saved filters, buyer and seller analytics, trust scores, profile completeness, traction metrics |
frontend-v2/is a Next.js App Router application using React 19 and a custom UI layer.- The product is organized into real workspace surfaces: marketplace, dashboard, sell studio, compare, favorites, messages, schedules, moderation, AI lab, policy, support, notifications, profiles, listings, and cart.
- Frontend state handles compare queues, notification inboxes, and app shell behavior while API traffic runs through a backend proxy route.
backend/src/server.tswires a modular Express 5 API with route groups for auth, listings, media, maps, cart, chat, dashboard, schedules, users, profiles, policies, AI, and admin features.- Request validation is handled with
zod, which gives the API a strong contract boundary. - Auth and authorization are enforced through middleware plus role checks, including admin-only paths.
- Supabase backs the application data model and operational views.
- The migration set covers marketplace foundation, holds and offers, notifications, saved filters, compare queues, media assets, AI diagnostics, meetup route persistence, and schedule confirmation state.
- The schema is not only feature-rich; it is hardened with row-level security and targeted indexes for common marketplace access patterns.
TomTompowers campus search, reverse geocoding, map tiles, and route planning.Chutes AIis used for listing assistance, moderation, and search refinement.- Both AI and maps are built with fallback logic so local development can still function when provider keys are absent.
- End-to-end product engineering across UI, API, database, and integration layers
- Modern frontend development with route-level UX, custom component systems, and role-aware workspaces
- Backend API design using modular Express routes, middleware, typed domain models, and schema validation
- Database design with migrations, policy enforcement, indexes, and analytics-friendly views
- Safety-aware system design through moderation, reporting, blocked-user logic, and thread controls
- AI integration done responsibly, with fallback behavior, observability, diagnostics, and persisted artifacts
- Geospatial product work through campus search, route computation, reverse geocoding, and meetup state persistence
- Document generation and export flows using PDF tooling for listings and moderation evidence packets
- Operational thinking through validation scripts, runbooks, startup helpers, and admin analytics
This repo supports draft creation, live publishing, relisting, duplication, archiving, sold state transitions, inventory tracking, media attachments, PDF export, and trust metadata. That is a far more realistic commerce lifecycle than a simple create-read-update-delete demo.
Messages are attached to listings and enriched with offers, hold states, and meetup context. The chat flow also screens for abusive language, off-platform contact pushes, private contact leaks, and risky behavior.
Meetups are not left as vague text in a chat thread. The app stores requested mode, time, location, route, confirmation state, and completion state, which makes the handoff workflow a real product feature.
Users can report listings, admins can apply moderation actions, evidence can be exported to PDF, and the platform exposes operator-facing analytics and AI diagnostics. This gives the system a real governance layer.
The AI layer does useful product work, but it is not a black box. The code records provider events, stores artifacts, exposes diagnostics, and degrades to fallback behavior when live providers are unavailable.
The migration history shows deliberate work around indexes, RLS policies, compare queue support, notification preferences, analytics trends, and access patterns for marketplace entities.
| Persona | Key surfaces |
|---|---|
| Buyer | Marketplace, compare, favorites, notifications, messages, schedules, listing detail, saved filters |
| Seller | Sell studio, profile workspace, listing management, traction analytics, conversations, meetup follow-through |
| Admin | Moderation desk, platform analytics, AI lab, enforcement actions, provider diagnostics, evidence export |
.
|-- backend/ Express + TypeScript API
|-- frontend-v2/ Next.js App Router frontend
|-- supabase/migrations/ Schema evolution, RLS, indexes, analytics views
|-- docs/ Runbooks, strategy notes, README assets
|-- scripts/ Setup and validation helpers
|-- start-mvp.cmd Local startup helper for the MVP workflow
|-- README.md Project showcase and setup guide
| Layer | Technology | Notes |
|---|---|---|
| Frontend | Next.js 16, React 19 | App Router, multi-page workspace UI |
| Styling | Vanilla CSS, custom primitives | Product-specific design system and interaction surfaces |
| Backend | Node.js, Express 5, TypeScript | Modular route groups and typed service layer |
| Validation | Zod | Runtime request validation and safer contracts |
| Data | Supabase, PostgreSQL | Marketplace data, profiles, filters, notifications, diagnostics |
| Maps | TomTom, Leaflet | Search, reverse geocode, campus route and tile support |
| AI | Chutes AI | Listing assist, moderation, search refinement |
| Document generation | pdf-lib | Listing and moderation evidence exports |
| Fetching | SWR plus custom API helpers | UI data loading and refresh flows |
| Tooling | tsx, TypeScript compiler, PowerShell helpers | Local development and workflow support |
npm installPowerShell:
Copy-Item .env.example .env
Copy-Item frontend-v2\.env.local.example frontend-v2\.env.localBash:
cp .env.example .env
cp frontend-v2/.env.local.example frontend-v2/.env.localnpm run dev:backendBackend default: http://127.0.0.1:4000
npm run dev:frontend:v2Frontend default: http://127.0.0.1:3001
npm run build.\start-mvp.cmdThis helper bootstraps the MVP workflow by building, selecting free ports, writing frontend env values, and launching both application surfaces.
- AI features can fall back to mock logic when
CHUTES_API_KEYis not configured. - Campus map search and walking routes can fall back to local campus-spot logic when
TOMTOM_API_KEYis absent. - The backend supports development auth bypass personas through
DEV_AUTH_BYPASS_ENABLEDandDEV_AUTH_BYPASS_EMAILS. - The frontend rewrites
/api/backend/*to the backend URL configured inUNIKART_BACKEND_URL.
Environment variables
| Variable | Purpose |
|---|---|
NODE_ENV |
Runtime mode |
PORT |
Backend port |
AUTH_OFF |
Disable auth gate for local scenarios |
SUPABASE_URL |
Supabase project URL |
SUPABASE_ANON_KEY |
Client-facing Supabase key |
SUPABASE_SERVICE_ROLE_KEY |
Admin access for server-side operations |
DEV_AUTH_BYPASS_ENABLED |
Enables demo-persona auth bypass |
DEV_AUTH_BYPASS_EMAILS |
Email list used for bypass personas |
RAZORPAY_KEY_ID |
Payment credential scaffold in env template |
RAZORPAY_KEY_SECRET |
Payment credential scaffold in env template |
CHUTES_API_KEY |
AI provider credential |
CHUTES_BASE_URL |
AI provider base URL |
CHUTES_COMPLETIONS_PATH |
AI completions endpoint path |
CHUTES_MODEL |
AI model identifier |
CHUTES_TIMEOUT_SECONDS |
AI timeout |
TOMTOM_API_KEY |
Maps credential |
TOMTOM_SEARCH_BASE_URL |
Search endpoint base |
TOMTOM_ROUTING_BASE_URL |
Routing endpoint base |
TOMTOM_MAPS_BASE_URL |
Tile endpoint base |
| Variable | Purpose |
|---|---|
NEXT_PUBLIC_API_BASE |
Browser-visible backend base |
UNIKART_BACKEND_URL |
Server-side backend target for rewrite/proxy usage |
NEXT_PUBLIC_SUPABASE_URL |
Frontend Supabase URL |
NEXT_PUBLIC_SUPABASE_ANON_KEY |
Frontend Supabase anon key |
Backend route groups
/health/auth/ai/admin/listings/media/maps/cart/chat/dashboard/schedules/users/profiles/policies
This project shows evidence of learning and implementation across:
- product scoping and feature decomposition
- multi-role UX design
- API modeling and validation
- relational data modeling
- row-level security and access control
- event and analytics thinking
- AI integration with safe fallbacks
- geospatial interaction design
- platform moderation workflows
- operational documentation and local tooling
That combination makes it a strong major-project candidate for a portfolio, internship discussion, academic showcase, or technical interview walkthrough.
CampusMarket is an actively evolving MVP codebase. The repository already demonstrates a broad and credible feature set for a campus marketplace platform, while still leaving obvious room for future polish around deployment, testing depth, payments, and production hardening.
- production deployment documentation and architecture notes
- end-to-end test coverage for critical flows
- richer seller analytics and marketplace insights
- payment flow completion and post-transaction dispute handling
- image optimization and media-processing pipeline upgrades
- deeper recommendation and search relevance features








