ApplyIQ

📃 Description

ApplyIQ is an app based on Perplexity Sonar API to scan the internet in real time for hidden job opportunities, match them to your skills and interests, and auto-generate tailored resumes and cover letters using deep research + reasoning.

🎯 Target users

• Job/Internship seekers — especially students, new grads, and career changers.

🧠 Why is this a good idea?

• Real problem: Job search is not efficient and time-consuming considering : scattered listings, generic resumes, and missed opportunities.
• Sonar advantage: Real-time search + reasoning + citations means accurate, up-to-date jobs and company insights.
• Unique twist: Combines internet search, research agents, and generative customization (all-in-one).
• Viral potential: People would love a “smart job search AI” that handles search + resume tailoring.

💻 Tech stack

• Backend : Python (FastAPI)
• Frontend : NextJS/ReactJS/TSX + Tailwind CSS (or Material UI)
• AI & Search API : Perplexity Sonar API
• Hosting : Vercel

👉🏻 Core Features / Use cases

• Feature 1: Smart Job Discovery Input: User enters job of interest Sonar API is used to: Search for latest jobs matching profile (with filters) Get company insights and role expectations using sonar-deep-research sonar-reasoning Show real-time results with citations -> links

• Feature 2 : Display the key skills and requirements that match the job searched Input: Upload basic resume or fill form Sonar Reasoning Pro Sonar Reasoning is used to : Generate requirements and skills keywords for each role looked for Data mining -> search made by scaning the keywords and making sense of all of it, serialize it then render it (multiple regex are used for this)

• Feature 3 : Estimated salary based on the region/remote For a selected displayed job, generate:

  • range of salary
  • future insights/promotions
  • sources

⚙️ Set up the project environment

Conda virtual environment (fastapi-env)

conda activate fastapi-env conda deactivate

FastAPI and uvicorn

pip install fastapi uvicorn

CORS support for API calls (browser based requests)

pip install "python-multipart" fastapi[all] || pip install fastapi[all]

Run the backend

uvicorn main:app --reload --port 8000 Runs on : http://127.0.0.1:8000 || http://localhost:8000

Run the frontend

npm run dev Runs on : http://127.0.0.1:3000 || http://localhost:3000

Frontend Installed packages

• App Router (Y)
• TypeScript (Y)
• ESLint (Y)
• Turbopack (Y)

Frontend Installed dependencies

Installing dependencies:

• react
• react-dom
• next

Installing devDependencies:

• typescript
• @types/node
• @types/react
• @types/react-dom
• @tailwindcss/postcss
• tailwindcss
• eslint
• eslint-config-next
• @eslint/eslintrc

Summary of Commands to Start Everything -- After the setup

Backend:

cd backend
conda activate fastapi-env
uvicorn main:app --reload --port 8000

Frontend (to be opened in another terminal):

cd frontend
npm run dev

More on Perplexity API (can also be found on the website under 'Getting Started')

Models used

• sonar-deep-research (too slow ❌)
• sonar-reasoning-pro ✅
• sonar (too basic ⚠️)

Roles used

• user
• assistant

Object of research

The goal of this project is to be able to fetch the AI API thoroughly to retrieve the information needed to locate the positions the user's looking for. To achieve this, the "custom research agents" concept will be efficient since it offers the following options :

• Multi-turn conversations -> user job specifications/conditions
• Advanced reasoning
• More control over model behavior (temperature, stream, etc.)
• Chain-of-thought generation (great for agents, tools, research bots) /chat/completion : "https://api.perplexity.ai/chat/completions"

💡 Further considerations (extensions)

• Add a Chrome extension to scan job boards like LinkedIn
• Track applications and send alerts when new jobs match

Author

• Kaoutar El Azzab

Demo Link

On youtube