This is the final handoff for continuing development of the Nodus Workforce Intelligence System.
- Frontend is fully built and deployed-ready (Vercel)
- Backend is fully built but NOT deployed yet (will be deployed on Railway)
- PostgreSQL database is fully configured and working
- AI integration layer is NOT built yet (THIS IS THE MAIN TASK)
Final system will run as:
Frontend (Vercel) ↓ Backend API (Railway) ↓ PostgreSQL Database ↓ OpenAI API (AI Intelligence Layer)
Frontend is fully built and production-ready.
- Landing page (interactive network visualization)
- Login / Register
- Job Seeker Dashboard
- Employer Dashboard
- Training Provider Dashboard
- Search Page (AI-ready UI already built)
- Directory Page (training / employers / talent)
- Profile Page (skills input system)
- History Page (AI results display UI)
✔ Fully functional UI
✔ Uses mock services (temporary only)
✔ Already designed to consume real backend APIs
Backend is fully implemented and tested locally.
- Login / registration
- Role-based access:
- JOB_SEEKER
- EMPLOYER
- TRAINING_PROVIDER
- Stored in PostgreSQL
- Job postings stored in DB
- Search API working
- Structured job data:
- title
- company
- location
- required skills
- timestamp (data freshness tracking)
✔ Verified via Postman
- Training programs stored
- Skills covered mapping
- Directory API implemented
- Employer job postings
- Talent search endpoint (mock logic only)
- User skills stored
- Target role stored
- Experience level stored
- Discoverability flag
- Endpoints exist
- PostgreSQL persistence works
- Recommendation objects stored correctly
✔ Postman tested
✔ PostgreSQL fully working
✔ Service/repository/controller layers stable
✔ No structural backend issues
This is the ONLY missing core component.
Currently:
- Recommendation system = mock data
- Skill gap analysis = NOT implemented
- Training alignment intelligence = NOT implemented
- Employer matching intelligence = NOT implemented
We will use OpenAI (or equivalent LLM API) to generate intelligence.
We will integrate OpenAI API key into Spring Boot.
IMPORTANT:
- Requires ~$5 minimum credit depending on OpenAI billing setup
- This is expected and fine for hackathon/demo usage
- Only inference calls (no training)
If OpenAI is not preferred:
- Anthropic Claude API
- Google Gemini API
- Any LLM API that supports structured responses
BUT: OpenAI is preferred for fastest Spring Boot integration.
Create:
AiRecommendationService
This service will:
- Collect backend data
- Build AI prompts
- Call OpenAI API
- Parse response
- Store result in PostgreSQL
Example inputs:
- User skills
- Selected job postings
- Training programs (optional)
Example (Job Seeker):
User Skills:
- Java
- SQL
- Spring Boot
Job Requirements:
- Java
- SQL
- Python
- AWS
TASK:
- Identify skill gaps
- Calculate match percentage
- Provide explanation
- Suggest learning path
- Recommend training programs
Send:
- system prompt (defines AI behavior)
- user prompt (data above)
Expected output:
{ "matchScore": 72, "missingSkills": ["Python", "AWS"], "recommendation": "Focus on Python first...", "trainingSuggestions": [ "AWS Bootcamp", "Python Data Engineering Course" ] }
Save:
- userId
- jobIds
- AI response JSON
- timestamp
This powers History page + analytics later
Frontend currently uses:
const USE_MOCK = true
You must:
- remove mock logic
- replace with real API calls to Railway backend
Target endpoint:
POST /api/recommendations/generate
User selects jobs → backend sends data → AI returns:
- skill gaps
- match score
- roadmap
Compare: curriculum vs job market demand
Output:
- missing skills
- alignment score
- curriculum gaps
Match: job requirements vs candidates
Output:
- ranked candidates
- skill overlap score
- pipeline readiness
CURRENT (MOCK):
Frontend → Mock Service → Fake Data
FINAL SYSTEM:
Frontend (Vercel) ↓ Spring Boot API (Railway — NOT DEPLOYED YET) ↓ PostgreSQL Database ↓ OpenAI API ↓ AI Response ↓ Frontend UI
Frontend is already AI-ready:
✔ Search Page:
- AI response panel already exists
✔ History Page:
- expects real AI results
✔ Profile Page:
- collects skills for AI input
✔ Dashboard Pages:
- already structured for AI insights
Transform system from:
static workforce data platform
TO
real-time AI workforce intelligence engine
- Build OpenAI integration in Spring Boot
- Create prompt engineering system
- Replace mock recommendation service
- Connect frontend → backend APIs
- Test full system end-to-end
System is complete when:
- User selects jobs
- AI analyzes skill gaps
- Returns explainable recommendation
- Saves to PostgreSQL
- History page displays AI output
- Fully working via Vercel + Railway
Everything except AI integration is complete.
This phase is ONLY:
Make the system intelligent using OpenAI.
Focus ONLY on:
- AI service layer
- prompt engineering
- API integration
- backend wiring
Do NOT modify frontend or backend architecture unless required for AI integration.