Career Ujjawalata

AI-powered career guidance for Indian students (Class 10 & 12)

A full-stack web application that recommends top 3 career paths based on a student's aptitude profile and subject interests — built with a Random Forest classifier, FastAPI backend, and an interactive single-page frontend.

---

Screenshots

Hero Screen	Assessment

Results	Career Details

---

Features

• 20-parameter assessment — 9 aptitude traits + 11 subject interest scores
• Likert scale UI — 5-option response scale (Strongly Disagree → Strongly Agree)
• Live radar charts — aptitude and interest profiles update in real time as you answer
• Top 3 career recommendations with confidence scores
• Career detail drawer — exams, top colleges, salary ROI, roles, study duration, competition level
• Expert connect modal — contact a domain expert for each career path
• Confetti reveal on results screen

---

Career Paths

Career	Key Signals
Engineering	High logic, maths, physics, computers
Medicine	High biology, chemistry, memory, empathy
Commerce / CA	High economics, business, numerical ability
Arts & Humanities	High verbal, creativity, literature, arts
Law	High verbal, logical reasoning, memory, history
Design & Architecture	High creativity, spatial ability, arts

---

Tech Stack

Layer	Technology
ML Model	scikit-learn RandomForestClassifier (200 trees, depth 12)
Data	Synthetic — Likert-sampled from archetype distributions
Backend	FastAPI + Pydantic v2
Server	Uvicorn (ASGI)
Frontend	Vanilla JS, Chart.js (radar), canvas-confetti
Styling	CSS custom properties, dark theme

---

Project Structure

career_guidance/
├── generate_data.py     # Synthetic dataset generator (Likert-sampled archetypes)
├── train_model.py       # Trains and saves Random Forest model
├── predict.py           # predict_top3() — importable inference function
├── app.py               # FastAPI app — serves UI + /predict endpoint
├── static/
│   └── index.html       # Full single-page wizard frontend
├── requirements.txt
└── README.md

---

Assessment Parameters

Aptitude (9 traits)

Parameter	Description
logical_reasoning	Deductive and inductive logic
numerical_ability	Arithmetic and quantitative aptitude
verbal_ability	Language comprehension and expression
creativity	Original thinking and ideation
analytical_thinking	Problem decomposition and analysis
leadership	Initiative, teamwork, and decision-making
spatial_ability	Visualising 2D/3D structures and relationships
memory_retention	Recall of facts, sequences, and concepts
empathy	Understanding and responding to others' needs

Subject Interests (11 subjects)

Physics · Chemistry · Biology · Mathematics · Literature · Economics · Arts · Computers · History · Psychology · Business

All parameters scored 0–100 (frontend maps Likert responses to 10 / 30 / 50 / 70 / 90).

API Reference

POST /predict

Accepts 20 numeric scores (0–100), returns top 3 career recommendations.

Request body

{
  "logical_reasoning": 90,
  "numerical_ability": 88,
  "verbal_ability": 52,
  "creativity": 58,
  "analytical_thinking": 87,
  "leadership": 62,
  "spatial_ability": 84,
  "memory_retention": 70,
  "empathy": 48,
  "interest_physics": 91,
  "interest_chemistry": 74,
  "interest_biology": 38,
  "interest_maths": 93,
  "interest_literature": 35,
  "interest_economics": 50,
  "interest_arts": 30,
  "interest_computers": 91,
  "interest_history": 33,
  "interest_psychology": 38,
  "interest_business": 52
}

Response

{
  "recommendations": [
    { "career": "Engineering",    "confidence": 94.5 },
    { "career": "Design Architecture", "confidence": 3.2 },
    { "career": "Commerce CA",    "confidence": 1.4 }
  ]
}

---

Model

• Algorithm: RandomForestClassifier — 200 estimators, max depth 12
• Training data: 1200 samples (200 per class), Likert-sampled from archetype (mean, std) distributions
• Features: 20 (9 aptitude + 11 interest)
• Classes: 6 career paths
• Train/test split: 80/20, stratified
• Accuracy: 99.58%

---