Development guidelines for building an anonymous identity-based job matching platform with XDMIQ credentialing, FastAPI backend, Next.js frontend, and multi-agent coordination system.
A comprehensive development skill for building an AI-powered, anonymous identity-based job matching platform with human-in-the-loop architecture and XDMIQ credentialing integration.
This skill guides development of a scalable job matching platform designed for 1B+ users, featuring:
1. **Human-in-the-Loop Priority**
- Always design AI decisions with human review queues
- Create state checkpoints for all AI operations
- Support rollback capabilities for AI-made decisions
- Log all AI decisions for audit and review
2. **Anonymous Identity Preservation**
- Never expose real user identity in any operation
- All authentication providers link to anonymous identity
- Maintain zero-knowledge architecture throughout
3. **Capability Over Credentials**
- Focus on demonstrated capabilities via XDMIQ scores
- Use preference-based questions for skill assessment
- Calculate and integrate XDMIQ scores (0-100) into matching
4. **State Recovery and Testing**
- Implement checkpoint systems for all critical operations
- Enable rollback of AI decisions
- Maintain audit logs for compliance and debugging
**Backend:**
**Frontend:**
**Backend (Python/FastAPI):**
**Frontend (TypeScript/Next.js):**
**Supported Providers:**
**Implementation Requirements:**
**Question Format:**
**Scoring:**
**When Implementing AI Features:**
1. **Design Human Review Queue**
- Create database table for pending AI decisions
- Implement admin interface for review
- Add approval/rejection workflows
2. **State Checkpointing**
- Save state before AI operation
- Store AI decision rationale
- Enable state restoration on rollback
3. **Error Handling**
- Implement OpenAI API with retry logic
- Add fallback strategies for API failures
- Log errors without exposing PII
4. **Audit Trail**
- Log all AI decisions with timestamps
- Store input context and output
- Enable compliance reporting
**Available Agent Categories:**
**C-Suite Agents:**
**Development Agents:**
**Project Management Agents:**
**Operations Agents:**
**Specialized Agents:**
**Agent Usage:**
**Step 1: Reference Existing Patterns**
**Step 2: Implement Business Logic**
**Step 3: Add Error Handling**
**Step 4: Test Anonymous Identity**
**Step 5: Implement Checkpoints**
**Target: 1B+ Users**
**Warp Terminal:**
**Claude Code:**
**Cursor IDE:**
**Example 1: Implementing AI-Powered Match**
```python
from fastapi import APIRouter, Depends
from sqlalchemy.orm import Session
from .models import MatchRequest, MatchResponse
from .services import ai_matching_service
from .checkpoint import create_checkpoint, restore_checkpoint
router = APIRouter()
@router.post("/match", response_model=MatchResponse)
async def create_match(
request: MatchRequest,
db: Session = Depends(get_db)
):
# Create checkpoint before AI operation
checkpoint_id = await create_checkpoint(
user_id=request.anonymous_id,
operation="match_creation",
state={"request": request.dict()}
)
try:
# AI matching (with human review queue)
match_result = await ai_matching_service.create_match(
anonymous_id=request.anonymous_id,
preferences=request.preferences,
xdmiq_score=request.xdmiq_score
)
# Queue for human review
await queue_for_review(
checkpoint_id=checkpoint_id,
match_result=match_result
)
return MatchResponse(
match_id=match_result.id,
status="pending_review",
anonymous_match_id=match_result.anonymous_match_id
)
except Exception as e:
# Restore checkpoint on failure
await restore_checkpoint(checkpoint_id)
raise
```
**Example 2: XDMIQ Assessment Component**
```typescript
// components/XDMIQAssessment.tsx
import { useState } from 'react';
import { submitAssessment } from '@/lib/api/xdmiq';
interface AssessmentProps {
anonymousId: string;
}
export function XDMIQAssessment({ anonymousId }: AssessmentProps) {
const [preference, setPreference] = useState('');
const [reasoning, setReasoning] = useState('');
const handleSubmit = async () => {
try {
const result = await submitAssessment({
anonymousId,
preference,
reasoning,
});
// Never expose real identity
console.log('Score updated for anonymous user:', result.score);
} catch (error) {
console.error('Assessment failed:', error);
}
};
return (
<div>
<h2>Which do you prefer?</h2>
<select value={preference} onChange={(e) => setPreference(e.target.value)}>
<option value="optionA">Option A</option>
<option value="optionB">Option B</option>
</select>
<h3>Why?</h3>
<textarea
value={reasoning}
onChange={(e) => setReasoning(e.target.value)}
placeholder="Explain your preference..."
/>
<button onClick={handleSubmit}>Submit Assessment</button>
</div>
);
}
```
**Example 3: Using Cursor Agents**
```
@fractional-ceo-agent Should we prioritize matching algorithm improvements or expand authentication providers?
@principal-dev-agent Review the anonymous identity implementation in auth/service.py
@project-manager-agent Create sprint plan for XDMIQ integration feature
@data-governance-agent Verify PII handling in matching algorithm
```
Leave a review
No reviews yet. Be the first to review this skill!
# Download SKILL.md from killerskills.ai/api/skills/ai-enabled-llc-matching-platform/raw