Pathlete

AI-powered career pathways for student-athletes. Explore non-playing careers tailored to your strengths and interests, generated with LLMs.

---

Overview

Pathlete helps college athletes discover realistic, high-potential career paths beyond playing professionally. The app collects a short survey of interests, major, and preferences; a backend Cloud Function synthesizes this context with an LLM to produce three prioritized career pathways with salary and education guidance. Results are displayed in a clean, mobile-friendly UI with follow-on curriculum suggestions.

Features

• Personalized survey using SurveyJS
• AI-generated, structured recommendations (3 ranked pathways)
• Salary and minimum education guidance per pathway
• Persisted quiz data and recommendations in Firestore
• React SPA with results and lesson plan views

Architecture

• Frontend: React (Create React App) in website/
• Backend: Firebase Cloud Functions (Node 18) in backend/functions/
• Database: Cloud Firestore
• LLM: OpenAI Chat Completions API

Data model (collections):

• quiz-data/{userId}: Raw survey submission
• recommendations/{userId}: AI output { pathways: [{ rank, name, avgSalary, minEduRequired }] }
• users/{userId}: Optional user profile (read by getUserInfo)

Cloud Functions (Callable):

• uploadQuizData: Stores quiz data for a userId, then triggers processing
• processQuizData: Calls OpenAI with context and saves recommendations
• getRecommendations: Fetches saved recommendations for a userId
• getUserInfo: Fetches user profile (if present)

Note: Frontend calls callable functions via raw HTTPS using Axios and the callable wire format. No Firebase client SDK is required in the browser for these calls.

Monorepo layout

upskilling/
  backend/
    firebase.json              # Functions codebase config
    functions/
      index.js                 # Cloud Functions (callable)
      package.json             # Node 18; axios, firebase-admin, firebase-functions
  website/
    package.json               # CRA scripts
    public/                    # Static assets
    src/                       # React app, Survey + Results views

Prerequisites

• Node.js 18.x
• npm 9+ (or pnpm/yarn)
• Firebase CLI (npm i -g firebase-tools) with access to a Firebase project
• OpenAI API key

Quick start (local dev)

1. Install dependencies

cd backend/functions && npm ci
cd ../../website && npm ci

2. Deploy or emulate the backend

• Fastest path: deploy callable functions and use them from local React

cd backend/functions
# Ensure your Firebase project is set
firebase use <your-project-id>

# Provide the OpenAI key to the runtime environment (see Environment for options)

# Deploy functions
npm run deploy

3. Configure frontend environment

Create website/.env.local with:

REACT_APP_UPLOAD_QUIZ_DATA_URL=https://<location>-<project-id>.cloudfunctions.net/uploadQuizData
REACT_APP_GET_RECOMMENDATIONS_URL=https://<location>-<project-id>.cloudfunctions.net/getRecommendations
REACT_APP_USER_ID=<a-stable-test-user-id>

4. Run the React app

cd website
npm start

The app will open at http://localhost:3000. Submitting the survey will hit your deployed callable functions.

Environment & configuration

Backend (Cloud Functions)

The code reads the OpenAI key from process.env.OPENAI_API_KEY.

You have two common options:

• Google Cloud Console: After first deploy, open your function > Edit > Runtime, build, connections > Runtime environment variables, add OPENAI_API_KEY.
• Local emulation: export the var before starting emulators.

export OPENAI_API_KEY=sk-...
cd backend/functions
npm run serve

Tip: If you prefer Firebase Config (functions.config()), update the code accordingly and set via firebase functions:config:set openai.api_key=....

Frontend (React)

The React app expects these variables at build/runtime:

• REACT_APP_UPLOAD_QUIZ_DATA_URL: Callable endpoint for uploadQuizData
• REACT_APP_GET_RECOMMENDATIONS_URL: Callable endpoint for getRecommendations
• REACT_APP_USER_ID: A user identifier string to associate survey data and results

Callable endpoint format:

• URL: https://<location>-<project-id>.cloudfunctions.net/<functionName>
• Method: POST
• Headers: Content-Type: application/json
• Body shape: { "data": { ... } }

Examples used in-app:

// Upload quiz data
POST <UPLOAD_QUIZ_DATA_URL>
{
  "data": {
    "userId": "<USER_ID>",
    "quizData": { /* survey payload */ }
  }
}

// Get recommendations
POST <GET_RECOMMENDATIONS_URL>
{
  "data": {
    "userId": "<USER_ID>"
  }
}

Deployment

Deploy Cloud Functions

cd backend/functions
firebase use <your-project-id>
npm run deploy

After deploy, copy callable URLs from the Firebase Console and place them in website/.env (or hosting provider env settings).

Deploy the React app

Build and deploy to your hosting provider of choice (Vercel, Netlify, Firebase Hosting, etc.).

cd website
npm run build
# Deploy the contents of website/build using your platform's CLI/UI

Firebase Hosting is not configured in this repo by default. You can add it via firebase init hosting in the repo root if you prefer a single Firebase deployment.

Troubleshooting

• JSON parsing errors when reading LLM output: reduce temperature, increase max_tokens, or enable JSON-formatted responses. The current backend prompts the model to return strict JSON and strips risky formatting, but vendor changes can still cause drift.
• 401/403 calling functions: ensure your callable endpoints are correct and that the function is deployed. If you later add auth/App Check enforcement, the Axios calls must include the appropriate tokens.
• OPENAI_API_KEY undefined: set it in the function runtime environment or export it before emulation.

Tech stack

• React 18, Create React App
• SurveyJS (survey-core, survey-react-ui)
• Firebase Cloud Functions v2, Node 18
• Cloud Firestore
• Axios
• OpenAI Chat Completions API

Credits

Texas Convergent EdTech Build Team

• Tech: Aastha Agrawal, Akshat Shah, Samuel Jacob
• Product: Ashley Chen, Aadit Jha, Tina Luo
• Design: Aashna Mangal