CredAI : Turn skills into proof

🚩 The Problem

When candidates apply for jobs, they list skills like Python, C++, SQL, or Machine Learning on their resumes.

But for recruiters, these skills are:

• Self-reported
• Unverified
• Often inflated or inaccurate

A resume saying "Proficient in Python debugging" does not prove that the candidate can actually debug Python code.

Recruiters either:

• Trust the resume blindly
• Or spend time validating skills through interviews and tests

This creates friction, bias, and inefficiency in hiring.

💡 The Solution — CredAI

CredAI replaces unverifiable resume skills with AI-evaluated, cryptographically verifiable micro-credentials.

Instead of claiming a skill, candidates can prove it.

CredAI:

• Runs a skill-specific AI oral exam
• Grades answers using a transparent rubric
• Produces evidence of reasoning and correctness
• Anchors a hash of that evidence on Solana
• Issues a verifiable credential ID that recruiters can independently check

🔍 What Makes CredAI Different

• Not a quiz — candidates must explain reasoning and debugging steps
• Skill-aware — exams adapt to the exact skill entered (free text)
• Verifiable — credentials are backed by an immutable blockchain proof
• Explainable — recruiters see scores, verdicts, and transcript evidence
• No trust required — verification does not rely on the issuer

🧠 How It Works (End-to-End)

1. Candidate enters any skill

Example: Python debugging, C++ OOP, SQL joins

2. AI oral exam begins

• Gemini generates a real, practical problem
• Candidate answers in text
• AI grades using a rubric (0–10)
• Adaptive follow-up questions are asked

3. Final evaluation

• If inconsistent → REVIEW
• If below threshold → FAIL
• If strong → PASS

4. Credential issuance (PASS only)

• Exam evidence is serialized into JSON
• A SHA-256 hash is computed
• Hash is written to Solana devnet via a Memo transaction
• A unique credentialId is issued

5. Recruiter verification

• Recruiter opens /verify
• Pastes credentialId
• Backend verifies:
  • Credential record
  • Solana transaction existence
  • Matching evidence hash
• Returns Verified ✅ with explorer link

📊 Admin Dashboard

CredAI includes an admin dashboard for transparency and auditability:

• Total sessions
• Pass / Fail / Review counts
• Issued credential count
• Flagged attempts

Per-session detail view:

• Full transcript (Q/A)
• Scores and feedback
• Final verdict
• Blockchain proof links

🛠 Tech Stack

Frontend

• Next.js (App Router)
• Tailwind CSS
• Client-side routing and state

Backend

• FastAPI (Python)
• Google Gemini 2.5 Flash (exam generation + grading)
• Solana devnet (credential proof anchoring)

Cryptography

• SHA-256 evidence hashing
• Solana Memo program for immutable storage

🔗 Key API Endpoints

Endpoint	Method	Description
/session/start	POST	Start exam with { "skill": "Python debugging" }
/session/{id}/answer	POST	Submit answer with { "answer": "..." }
/session/{id}/finalize	POST	Finalize exam
/credential/{credentialId}/verify	GET	Verify credential
/dashboard/metrics	GET	Get dashboard metrics
/dashboard/sessions	GET	Get all sessions
/dashboard/session/{id}	GET	Get session details

▶️ Running Locally

Backend

cd services/api_py
python -m venv .venv
.\.venv\Scripts\activate
pip install -r requirements.txt
uvicorn main:app --reload --port 8080

Frontend

cd apps/web_dashboard
npm install
npm run dev

Open: http://localhost:3000

🔐 Environment Variables

Backend (services/api_py/.env)

GEMINI_API_KEY=your_key_here
SOLANA_ISSUER_PRIVATE_KEY_BASE58=your_devnet_wallet_key

Frontend (apps/web_dashboard/.env.local)

NEXT_PUBLIC_API_BASE=http://localhost:8080

⚠️ Current Limitations

• In-memory storage (MVP)
• Solana devnet only
• No user authentication (candidate flow is open)

🚀 Future Improvements

• Persistent storage (SQLite / Firebase)
• PDF certificates with QR verification
• Auth-protected recruiter dashboard
• Multi-attempt skill history per candidate
• On-chain credential registry

🏁 Why CredAI Matters

Resumes list claims.

CredAI provides proof.

By transforming skills from unverifiable text into auditable, verifiable credentials, CredAI bridges the trust gap between candidates and recruiters — and makes skill-based hiring fairer, faster, and more transparent.