A Next.js application that helps users create personalized learning plans using the Vercel AI SDK. The app supports both OpenAI and Cerebras inference backends.

1. Install dependencies:

   npm install

2. Configure your AI provider:

   • Create a .env.local file with the provider selection and credentials. By default, the app will use OpenAI.

   # Select the provider: openai (default) or cerebras
   AI_PROVIDER=openai
   
   # OpenAI credentials (required when AI_PROVIDER=openai)
   OPENAI_API_KEY=your_openai_api_key
   
   # Cerebras credentials (required when AI_PROVIDER=cerebras)
   CEREBRAS_API_KEY=your_cerebras_api_key
   # Optional: override the base URL if you are using a private deployment
   # CEREBRAS_BASE_URL=https://api.cerebras.ai/v1

   • Modify lib/ai/config.ts to adjust the default models for each provider/use case. For example, gpt-oss-120b for Cerebras chat or gpt-5-mini for OpenAI.

3. Set up Supabase (required for persistence): Add these variables to .env.local:

NEXT_PUBLIC_SUPABASE_URL=https://your-supabase-project-ref.supabase.co
NEXT_PUBLIC_SUPABASE_ANON_KEY=your_public_anon_key_here
SUPABASE_SECRET_KEY=your_secret_key_here
SUPABASE_DB_URL=postgresql://postgres:[password]@db.[project-ref].supabase.co:5432/postgres

• The NEXT_PUBLIC_* keys are safe for the browser.
• SUPABASE_SECRET_KEY (a.k.a. service role key) and SUPABASE_DB_URL are server-only values; keep them out of the client bundle and version control. Rotate them immediately if they leak.

4. Run database migrations (after setting env vars):

   npm run db:generate
   npm run db:migrate

5. Run the development server:

   npm run dev

6. Run the background worker

   Course generation now happens fully asynchronously. There is a long-running worker that polls course_generation_jobs, calls the LLM, and persists results.

   Local development

   1. Make sure the required env vars are present (the easiest way is to export them in the shell that runs the worker; .env.local is not loaded automatically by tsx):

      export SUPABASE_DB_URL=postgresql://postgres:[password]@db.[project-ref].supabase.co:5432/postgres
      export OPENAI_API_KEY=your_openai_api_key
      export AI_PROVIDER=openai
      # Optional tuning
      export COURSE_GENERATION_WORKER_CONCURRENCY=3
      export COURSE_GENERATION_HEARTBEAT_MS=45000
      export COURSE_GENERATION_STALE_TIMEOUT_MS=180000
      export COURSE_GENERATION_REQUEUE_INTERVAL_MS=60000

      You can add these to your shell profile or use direnv/dotenvx to load them automatically.

   2. Start the worker in a separate terminal:

      npm run worker:course

      This script uses tsx to run worker/course-generator.ts directly. Keep it running alongside npm run dev.

   Render (production)

   1. Create a Render Background Worker service pointing to this repo/branch.

   2. Set the same environment variables in the Render dashboard (SUPABASE_DB_URL, OPENAI_API_KEY, AI_PROVIDER, plus any optional tuning vars).

   3. Use a start command that installs deps (with dev deps, for tsx) and launches the worker. For example:

      npm install
      npm run worker:course

      Alternatively you can build once and run the compiled JS with node dist/worker/course-generator.js if you prefer not to ship dev deps.

7. Open your browser: Navigate to http://localhost:3000