Full Stack Developer Internship - Demo
Task
🎯 Objective
Build a production-grade, end-to-end system.
You must choose ONE track only:
● Track A → AI App Generator (Hard / High Signal)
● Track B → College Discovery Platform (Product Build)
● Track C- Compensation intelligence system
👉 Your selection will NOT affect evaluation negatively.
👉 We are evaluating depth, ownership, and execution quality within your chosen track.
🛣️ Choose Your Track
🧠 Track A: AI App Generator (Advanced /
Systems Thinking)
Build a system that converts structured configuration (JSON) into a fully working, deployed
web application (frontend + backend + database).
This should behave like a mini app generator, not a single hardcoded app.
Reference - [Link]
Submission Link - [Link]
🧩 Problem Statement
You will work with structured configurations (JSON) that define:
● UI (forms, dashboards, tables)
● APIs (endpoints, actions)
● Database structure
● Authentication
These configurations may be:
● incomplete
● inconsistent
● partially incorrect
Your system must still work reliably under these conditions.
🚀 What You Need to Build
1. Dynamic Application Runtime (Core)
Build a system that:
● Reads JSON configuration
● Dynamically generates:
○ UI (forms, tables, dashboards)
○ Backend APIs
○ Database structure
⚠️ This must NOT be hardcoded.
2. Frontend
● Use [Link] ([Link] preferred)
● Render UI dynamically from config
Handle:
● missing fields
● unknown components
Include:
● loading states
● error states
3. Backend
● Use [Link] (TypeScript preferred)
● Generate APIs dynamically
Implement:
● validation layer
● error handling
● CRUD operations
4. Database
● Use PostgreSQL (preferred)
● Store and retrieve application data
Handle:
● optional fields
● schema mismatches
5. Authentication
● Implement basic auth (email/password or equivalent)
● Ensure user-scoped data access
6. Extensibility (Important)
Your system should allow:
● adding new UI components easily
● extending features without rewriting core logic
⚠️ Rigid systems will be considered weak.
7. Feature Implementation (Build ANY 3 — Mandatory)
You must implement at least 3 features from the list below.
These should be:
● fully functional (end-to-end)
● integrated into your system (not isolated hacks)
● working with your config-driven architecture
🧩 Platform & App Capabilities
● Multi-language / Localization
○ Multi-language UI support
○ Dynamic switching via config or user
● Mobile-ready experience
○ Fully responsive apps
○ (Bonus: PWA / installable)
🔌 Integrations & Ecosystem
● Export full codebase to GitHub
○ Generate project structure
○ Push to repository
● CSV Import System
○ Upload → map → store → render
● Post-creation integrations
○ Ability to add features after app is generated
🔐 Authentication & User Management
● Multiple login methods
○ Email/password + one additional method
● Customizable auth UI
○ Config-driven login/signup
● User-scoped data access
🔔 Notifications & Communication
● Event-based notifications
○ Trigger on user/system actions
● Email system (basic)
○ Transactional emails (mock acceptable)
⚠️ Feature Expectations
● Must work within your system, not standalone
● Should be config-driven where possible
● Must handle edge cases and failures
8. Deployment (Mandatory)
● Deploy a fully working system
● Ensure end-to-end usability
🧪 What We Will Test
We will:
● modify your configuration
● add/remove fields
● introduce edge cases
Your system should not break under these conditions.
📤 Submission (Mandatory)
Submit via Google Form (link will be provided):
● Live URL (working product)
● GitHub repository
● Loom video (5-10 min) explaining:
○ architecture
○ decisions taken
○ edge cases handled
○ tradeoffs
⚠️ Important Instructions
● You are expected to figure things out independently
● You may use any tools (Cursor, ChatGPT, Claude, etc.)
● Focus on building a working system, not perfect code
● If your system breaks under real usage, it will be rejected
🧠 Evaluation Focus
We are evaluating your ability to:
● build reliable systems from imperfect inputs
● design for extensibility and change
● handle real-world edge cases
● ship a complete, working product
● implement deep, integrated features (not superficial)
🏗️ Track B: College Discovery Platform
(Product Execution)
Inspired by: [Link] [Link]
🎯 Objective
Build a production-grade MVP of a college discovery + decision platform.
You are NOT required to build everything.
👉 You MUST:
● Review the feature list below
● Pick ANY 3–4 features
● Build them end-to-end (frontend + backend + DB + deploy)
🧩 Feature List (Choose 3–4)
1. 🔍 College Listing + Search
Build a listing page with:
● College cards:
○ Name
○ Location
○ Fees
○ Rating
● Search by college name
● Filter (any 2):
○ Location
○ Fees
○ Course
👉 Expectation:
● Fast, usable UI (not static dump)
● Pagination or infinite scroll
2. 🏫 College Detail Page
Build a detail page with:
● Overview:
○ Fees
○ Courses offered
○ Basic info
● Sections (at least 2):
○ Courses
○ Placements
○ Reviews (mock allowed)
👉 Expectation:
● Clean layout
● Proper routing ([Link] preferred)
3. ⚖️ Compare Colleges (HIGH PRIORITY FEATURE)
Build:
● Select 2–3 colleges
● Show comparison table:
○ Fees
○ Placement %
○ Rating
○ Location
👉 This is a decision feature, not UI feature
4. 🧠 Simple Predictor Tool
Build:
● Input:
○ Exam (e.g., JEE)
○ Rank
● Output:
○ List of colleges
👉 Logic:
● Can be rule-based or dataset-driven
5. 💬 Q&A / Discussion
Build:
● Ask question
● Answer question
● List view of questions
👉 Keep it simple but functional
6. 🔐 Auth + Saved Items
Build:
● Login system
● Save:
○ Colleges
○ Comparisons
⚠️ What You MUST Do
1. End-to-End Build
● Frontend + Backend + Database
● No partial work
2. Real Data Handling
● Use:
○ Mock dataset OR AI-generated dataset
● But:
○ Store in DB
○ No hardcoding everything in UI
3. Deployment (MANDATORY)
● Frontend → Vercel (or similar)
● Backend → Railway / Render (or similar)
● Live working URL required
🧱 Tech Requirements (Strict)
Frontend:
● [Link] or [Link]
● Tailwind CSS
Backend:
● [Link] (TypeScript preferred)
● REST APIs
Database:
● PostgreSQL (preferred)
🤖 AI Usage
You are encouraged to use:
● OpenAI
● Anthropic
● Cursor, ChatGPT, etc.
TRACK 3
Build a Compensation Intelligence System (India →
Global)
🎯 CONTEXT
Build a production-grade compensation intelligence system inspired by:
● [Link] (primary reference)
● 6figr
● AmbitionBox
● Glassdoor
● IndiaTechSalaries
🧠 CORE CONCEPT (NON-NEGOTIABLE)
Levels > Titles
● L3 / L4 / L5 ≠ SDE1 / SDE2
● Same title ≠ same compensation
● Compensation is tied to level standardization
👉 Why [Link] works
👉 Where AmbitionBox / Glassdoor fail
🧠 PRODUCT PRINCIPLE
Structured → Comparable → Decision-ready
You are building:
● ✅ A compensation intelligence system
You are NOT building:
● ❌ A salary listing site
🧠 PHASE 1: RESEARCH (MANDATORY)
Reverse Engineer Platforms
Analyze:
● Salary listing pages
● Company pages
● Submission flows
● Comparison features
Output (Required)
● Observations
● What works / fails
● Key differences
● Gaps
❌ No screenshots
✅ Structured insights only
Feature Mapping Sheet
| Feature | [Link] | 6figr | AmbitionBox | Glassdoor | Build? |
Rules:
● Be opinionated
● Only mark YES for features you will build
● Focus on core system
🎨 Creative Freedom (10–20%)
Allowed:
● Improve filtering UX
● Simplify flows
● Add clarity
Not allowed:
● New product categories
● Side features
🛠️ PHASE 2: BUILD (3 DAYS)
🎯 Mission
Build a system that is:
Structured → Queryable → Comparable → Decision-ready
🧱 CORE SYSTEM
DATABASE (PostgreSQL + Prisma)
Salary Table
● company (normalized)
● role
● level (L3/L4/L5)
● location
● experience_years
● base_salary
● bonus
● stock
● total_compensation
● confidence_score
Rules
● total = base + bonus + stock
● bonus/stock default = 0
● normalize company names (lowercase, trimmed)
⚙️ BACKEND APIs
POST /ingest-salary
● Validate input (strict)
● Normalize company
● Compute total
● Store
GET /salaries (CORE API)
● Filters: company, role, level, location
● Sorting: total_compensation
● Server-side filtering
GET /company/:company
● Salary list
● Median compensation
● Level distribution
GET /compare
● Input: 2 salary IDs
● Output:
○ base
○ bonus
○ stock
○ total
○ level difference
🖥️ FRONTEND (MANDATORY PAGES)
Use React / [Link] + Tailwind
1. HOME PAGE
● Clear positioning (levels-based compensation insights)
● Search / filter entry (company / role / level)
● Navigation to:
○ Salary Table
○ Company Page
○ Compare Page
👉 Functional, not marketing-heavy
2. SALARY TABLE PAGE (CORE)
Columns:
● Company
● Role
● Level
● Location
● Experience
● Total Compensation
Features:
● Filters (company, role, level, location)
● API-driven table
● Sorting
● Empty states
3. COMPANY PAGE
● Salary list
● Median compensation
● Level distribution
4. COMPARE PAGE
● Select 2 salaries
● Structured comparison:
○ base
○ bonus
○ stock
○ total
○ level difference
⚠️ EDGE CASES (MANDATORY)
● Missing bonus/stock → 0
● Duplicate entries
● Company variations (“Google”, “ google ”, “GOOGLE”)
● Invalid data → reject
🔄 DATA CONTRACT
{
company,
role,
level_standardized,
base_salary,
bonus,
stock,
confidence
Rules:
● Revalidate everything
● Never trust input blindly
🚫 DO NOT BUILD
● Auth
● Reviews
● Chat
📤 Submission (Mandatory)
Submit via Google Form (link will be provided):
● Live URL (working product)
● GitHub repository
● Loom video (5-10 min) explaining:
○ architecture
○ decisions taken
○ edge cases handled
○ tradeoffs
⚠️ Important Note
This task is intentionally challenging.
We have designed it to create a steep learning curve and to identify people who can
operate with high ownership and high agency.
You are not expected to know everything beforehand.
But you are expected to:
● figure things out independently
● navigate ambiguity
● make decisions without waiting for instructions
● push through technical and product challenges
This is not a typical internship assignment.
It reflects the kind of problems you will work on here.
If you’re someone who enjoys:
● building from scratch
● solving unclear problems
● taking full ownership
you will find this exciting.
If you prefer structured, step-by-step tasks, this may not be the right fit.