CareerCraft

An AI-powered career assistant for resumes, ATS optimization, and interview preparation.

CareerCraft is a full-stack web application built to solve a real problem faced by students and early-career professionals: understanding how resumes are evaluated, how ATS systems work, and how to improve job readiness using AI in a practical, explainable way.

This project prioritizes clarity, correctness, testing, and real-world usefulness over flashy or opaque AI features.

---

Why CareerCraft?

Many applicants struggle with:

• Low ATS scores without knowing why
• Overwhelming job descriptions
• Unclear skill expectations
• Weak resume bullet points
• Inefficient interview preparation

CareerCraft was created to:

• Explain why a resume matches or doesn't
• Provide actionable, prioritized feedback
• Use AI responsibly and transparently
• Be beginner-friendly while technically sound

---

What the App Does

Resume & ATS Analysis

• Computes ATS match score (0–100)
• Compares resume and job description meaningfully
• Safely handles empty, short, or invalid inputs

Skill Intelligence

• Identifies matched skills
• Detects missing skills
• Calculates skill match percentage
• Recommends high-impact skills to learn next

Career Preparation

• Generates tailored cover letters
• Improves resume bullet points
• Creates interview questions from job descriptions
• Provides career advice based on resume + role

Resume Utilities

• Parses PDF and DOCX resumes
• OCR fallback for scanned documents
• Extracts clean, usable text

Management

• Tracks user history of actions and analyses

---

Tech Stack

Frontend

• Angular (standalone components)
• TypeScript
• RxJS
• Karma + Jasmine for unit testing
• Cypress for E2E testing

Backend

• FastAPI
• Python 3.13
• scikit-learn
• Sentence Transformers
• TF-IDF + cosine similarity
• PyPDF2, python-docx, pytesseract

---

Project Structure

careercraft/
├── backend/
│   ├── app/
│   │   ├── routers/
│   │   ├── services/
│   │   └── main.py
│   ├── tests/
│   │   ├── test_ats.py
│   │   ├── test_job_fit.py
│   │   ├── test_skill_gap.py
│   │   ├── test_recommend_skills.py
│   │   ├── test_resume_optimizer.py
│   │   ├── test_resume_parser.py
│   │   └── test_cover_letter.py
│   └── requirements.txt
│
├── frontend/
│   └── careercraft/
│       ├── src/app/pages/
│       ├── src/app/service/
│       ├── src/app/app.routes.ts
│       └── src/app/app.spec.ts
│
├── cypress/
│   ├── e2e/
│   │   ├── home.cy.js
│   │   ├── ats-score.cy.js
│   │   ├── cover-letter.cy.js
│   │   ├── job-fit.cy.js
│   │   ├── skill-gap.cy.js
│   │   ├── skill-recommendation.cy.js
│   │   ├── career-advice.cy.js
│   │   ├── interview-questions.cy.js
│   │   ├── improve-bullet.cy.js
│   │   ├── resume-parser.cy.js
│   │   └── history.cy.js
│   └── support/
│       ├── commands.js
│       └── e2e.js
├── cypress.config.js
└── README.md

---

How to Run the Project

Backend

cd backend
python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txt
uvicorn app.main:app --reload

Backend runs at: http://127.0.0.1:8000

---

Frontend

cd frontend/careercraft
npm install
ng serve

Frontend runs at: http://localhost:4200

---

Testing Philosophy

CareerCraft follows a realistic, user-focused testing approach across three layers:

Layer	Framework	Focus
Backend unit tests	Pytest	Correctness, edge cases, safety
Frontend unit tests	Karma + Jasmine	Component logic, state, async
E2E tests	Cypress	Full user flows, API integration

Core principles:

• Test what users can break
• Test edge cases before happy paths
• Avoid meaningless "always green" tests
• Each layer tests what it is best suited for

---

Backend Tests (Pytest)

Backend tests focus on correctness, safety, and edge-case handling across all API endpoints.

Test Files

File	Endpoint	What Is Tested
test_ats.py	/ai/ats-score	Valid input, empty inputs, short resume, long input, non-technical JD, score bounds
test_job_fit.py	/ai/job-fit	Valid input, empty inputs, both empty, long input, score bounds
test_skill_gap.py	/ai/smart-skill-gap	Matched/missing skills, empty inputs, full match, no skills in JD
test_recommend_skills.py	/ai/recommend-skills	Recommendations returned, no skills in JD, all skills matched
test_resume_optimizer.py	/ai/resume-optimizer	Valid input, empty resume, empty job description
test_resume_parser.py	/ai/parse-resume-advanced	Invalid file type, empty file
test_cover_letter.py	/llm/cover-letter	Valid input, empty inputs, long input, non-technical JD, response format

Why This Matters

Real ATS systems often fail silently.
These tests ensure no crashes, no invalid scores, and predictable behavior under all conditions.

Run Backend Tests

cd backend
.venv\Scripts\activate

# Run all tests
python -m pytest -v

# Run a single file
python -m pytest tests/test_ats.py -v

---

Frontend Unit Tests (Karma + Jasmine)

Frontend unit tests focus on component logic and UI behavior in isolation.

What Is Tested

• Component creation
• Input validation (empty fields)
• API calls triggered only when inputs are valid
• UI updates after API responses
• Error handling without crashes
• Standalone component behavior (no NgModules)

Run Frontend Unit Tests

cd frontend/careercraft
npm test

---

E2E Tests (Cypress)

E2E tests cover full user flows from the browser through to the API, testing every page of the application.

E2E Coverage

Page	What Is Tested
Home	Layout, navigation to all pages
ATS Score	Valid input, loading state, button disabled, error handling
Cover Letter	Valid input, loading state, button disabled, error handling
Job Fit	Valid input, loading state, button disabled, error handling
Skill Gap	Valid input, loading state, clear button, empty results
Skill Recommendations	Valid input, loading state, no recommendations state
Career Advice	Valid input, loading state, partial input validation
Interview Questions	Valid input, loading state, empty results state
Improve Bullet	Valid input, loading state, button disabled, error handling
Resume Parser	PDF/DOCX upload, loading state, unsupported file type, error handling
History	Empty state, populated after completed analysis

E2E Test Patterns

• HTTP interception — all API calls intercepted with cy.intercept to control responses
• Loading state assertions — button disabled and loading message visible during requests
• Network delay simulation — delay property forces async states to be testable without hitting the real backend
• Error handling — 500 responses verified to trigger correct user-facing alerts
• Input validation — empty and partial inputs verified to block submission
• Edge cases — unsupported file types, empty API result arrays, zero scores

Run E2E Tests

Requires both the frontend and backend to be running first.

# Open Cypress interactive UI
npx cypress open

# Run all tests headless
npx cypress run

# Run a single spec
npx cypress run --spec "cypress/e2e/ats-score.cy.js"

---

Pages Implemented

• Home Page
• ATS Score Page
• Career Advice Page
• Job Fit Page
• Skill Gap Analysis Page
• Skill Recommendation Page
• Cover Letter Generator
• Interview Questions Generator
• Resume Parser
• Resume Bullet Improvement
• History Page

All pages include input validation, loading states, error handling, and consistent UX patterns.

---

Key Design Decisions

• Standalone Angular components for cleaner architecture
• Service-based API abstraction
• ML logic isolated in backend services
• No silent failures
• Human-readable AI output

---

Future Improvements

• User authentication
• Persistent history storage (database)
• Export reports as PDF
• Skill learning roadmaps
• Cloud resume uploads
• Cypress tests integrated into CI/CD pipeline