SmartScreen - AI-Driven Screening and Validation Platform

An AI-powered platform connecting students and recruiters through intelligent skill validation, fair evaluation, and structured recruitment workflows.

📋 Project Overview

SmartScreen is a two-stage platform that combines developer learning with AI-based recruitment:

1. Learning Stage (SkillForge) - Students practice real-world engineering scenarios, improve skills, and build portfolios
2. Recruitment Stage (AI Screening) - Students apply to jobs, undergo AI voice interviews, and recruiters make data-driven decisions

The platform eliminates resume bias, prevents coding test cheating through reasoning-based interviews, and creates a fair, skill-based evaluation ecosystem.

🏗️ System Architecture

flowchart TD
    subgraph INPUT["📊 INPUT LAYER"]
        I1["👨‍💻 Student<br>Profile"] 
        I2["📝 Resume &<br>GitHub"]
        I3["🎤 Skill<br>Data"]
    end

    subgraph LEARNING["🎓 LEARNING PATH"]
        L1["📚 Practice<br>Tasks"]
    end

    subgraph JOB["💼 JOB POSTING"]
        J1["🏢 Recruiter<br>Posts Job"]
    end

    subgraph DATA["⚙️ DATA PROCESSING"]
        D1["Profile<br>Analysis"]
        D2["Feature<br>Extraction"]
    end

    subgraph AI_STAGE["🤖 AI EVALUATION LAYER"]
        A1["🧠 Technical<br>Reasoning"]
        A2["💬 Communication<br>Clarity"]
        A3["🔧 Problem<br>Solving"]
    end

    subgraph AGGREGATE["📊 AGGREGATION"]
        A4["✨ Score<br>Combiner"]
    end

    subgraph OUTPUT["📈 OUTPUT LAYER"]
        O1["📋 Readiness<br>Report"]
        O2["👔 Recruiter<br>Decision"]
    end

    I1 & I2 & I3 --> D1
    L1 & J1 --> D1
    D1 --> D2
    D2 --> A1
    D2 --> A2
    D2 --> A3
    A1 --> A4
    A2 --> A4
    A3 --> A4
    A4 --> O1 & O2

    style INPUT fill:#e8f5e9,stroke:#2e7d32,stroke-width:2px,color:#1b5e20
    style LEARNING fill:#fff9c4,stroke:#f57f17,stroke-width:2px,color:#bf360c
    style JOB fill:#f3e5f5,stroke:#6a1b9a,stroke-width:2px,color:#4a148c
    style DATA fill:#e3f2fd,stroke:#1565c0,stroke-width:2px,color:#0d47a1
    style AI_STAGE fill:#fff3e0,stroke:#e65100,stroke-width:2px,color:#bf360c
    style AGGREGATE fill:#ffe0b2,stroke:#e65100,stroke-width:2px,color:#bf360c
    style OUTPUT fill:#c8e6c9,stroke:#2e7d32,stroke-width:2px,color:#1b5e20

Loading

🔄 Complete Workflow

Stage 1: Learning Phase

• Students practice real-world tech tasks (system design, debugging, API development)
• AI provides personalized feedback on code quality and reasoning
• Track skill readiness score across multiple dimensions
• Build authentic portfolio projects

Stage 2: Recruitment Phase

Step 1: Job Posting

• Recruiters post jobs with:
  • Job title & description
  • Required tech stack
  • Experience level
  • Role type (Backend/Frontend/ML/etc.)
• Job visible to all students

Step 2: Student Application

• Students apply by submitting:
  • Resume
  • GitHub profile link
  • LeetCode profile
  • Platform skill score
• Application stored with candidate signals

Step 3: Round 1 - AI Screening Interview

The AI conducts a structured voice interview evaluating:

Technical Reasoning

• Why would you use Redis instead of PostgreSQL?
• How would you scale an API to 1M users?
• Explain your system design approach

Debugging Thinking

• Your API latency increased suddenly. What do you check first?
• How would you approach debugging this issue?

Communication Clarity

• Analyze speech patterns, explanation structure, confidence
• Measure how well candidate articulates technical concepts

Problem Decomposition

• Does candidate break problems into manageable parts?
• Can they explain step-by-step approach?

Step 4: AI Evaluation Report

The system generates a structured report:

Candidate: John Doe

Communication Score: 8.2 / 10
Technical Reasoning Score: 7.6 / 10
Problem Solving Score: 8.0 / 10
System Design Awareness: 7.4 / 10

Overall Readiness Score: 7.8 / 10
Recommendation: Proceed to Round 2 Interview

Step 5: Round 2 - Recruiter Decision

• Recruiter receives AI-evaluated candidates
• Makes final hiring decision based on structured report
• Recruiter does NOT browse resumes (removes bias)
• Focus on demonstrated ability rankings

✨ Key Features

For Students

• ✅ Real-time skill practice with AI guidance
• ✅ Personalized learning recommendations
• ✅ Portfolio building on the platform
• ✅ Readiness score tracking
• ✅ AI mock interview practice
• ✅ Fair, transparent skill evaluation
• ✅ Direct job application with verified skills

For Recruiters

• ✅ Automated candidate screening (Round 1)
• ✅ Structured evaluation reports
• ✅ Pre-screened, qualified candidates
• ✅ 60-70% reduction in screening time
• ✅ Elimination of resume bias
• ✅ Communication + reasoning evaluation
• ✅ Transparent, defensible hiring decisions

🤖 AI Agent Architecture

Agent 1: Task Generator

• Generates interview questions based on job role
• Tailored to required tech stack
• Difficulty scaled to experience level

Agent 2: Technical Evaluator

• Analyzes logical reasoning and correctness
• Measures depth of technical understanding
• Outputs technical reasoning score

Agent 3: Communication Evaluator

• Analyzes clarity, structure, confidence
• Measures articulation quality
• Outputs communication score

Agent 4: Score Aggregator

• Combines all evaluation metrics
• Generates final readiness score
• Produces recruiter recommendation

📊 Evaluation Metrics

Each candidate receives scores in these categories:

Metric	What It Measures
Communication Score	Clarity of explanation, articulation quality
Technical Reasoning Score	Depth of understanding, logical thinking
Problem Solving Score	Ability to decompose problems, systematic approach
System Design Awareness	Architectural thinking, scalability considerations
Overall Readiness Score	Combined performance metric for hiring decision

💡 Why This Approach Works

vs. Traditional Resume Screening

• ❌ Resume bias (school prestige, formatting, name bias)
• ✅ SmartScreen: Ability-based evaluation only

vs. Coding Tests (LeetCode, HackerRank)

• ❌ Easy to cheat with ChatGPT (78% of problems solvable by GPT-4)
• ✅ SmartScreen: Voice interviews test reasoning, not memorization

vs. Existing AI Hiring (HireVue, Modern Hire)

• ❌ Students apply cold without preparation
• ✅ SmartScreen: Students learn first, then apply prepared

🛠️ Tech Stack

Frontend

• React / Next.js
• TailwindCSS
• shadcn/ui

Backend

• Node.js + Express OR Python FastAPI
• PostgreSQL / MongoDB
• Redis (for caching)

AI & NLP

• OpenAI API (GPT-4 for evaluation)
• Whisper API (speech-to-text)
• Custom evaluation prompts

Infrastructure

• Docker / Kubernetes
• AWS / GCP deployment
• CI/CD pipeline

📦 Database Schema

Users Table

• id, name, email, password, role (student/recruiter)

Student Profiles

• user_id, github_url, leetcode_url, resume_url, skills, readiness_score

Recruiter Profiles

• user_id, company_name, company_description

Jobs Table

• job_id, recruiter_id, title, description, required_skills, experience_level

Applications Table

• application_id, student_id, job_id, status, evaluation_score, report

Interview Records

• interview_id, application_id, technical_score, communication_score, overall_score, ai_recommendation

🚀 MVP Scope

For initial college capstone and demo:

• User authentication (Student + Recruiter)
• Student profile creation
• Recruiter job posting
• Job application system
• AI evaluation of responses
• Structured recruiter report
• Full coding challenge environment
• Automated GitHub analysis
• Advanced analytics dashboard

🔐 Security & Fairness

• Password hashing (bcrypt)
• Input validation and sanitization
• API rate limiting
• Secure file uploads
• Bias monitoring and auditing
• Transparent AI decision logs
• Human-in-the-loop final decisions

📈 Success Metrics

The platform is successful if it:

• ✅ Evaluates developer reasoning accurately
• ✅ Reduces recruiter screening time by 60-70%
• ✅ Helps students practice real interview scenarios
• ✅ Increases fair hiring outcomes (reduces bias)
• ✅ Achieves 78%+ correlation between AI scores and actual job performance

🎯 Future Enhancements

• Full coding challenge environment with automated testing
• Real-time collaborative coding sessions
• Skill progression tracking and visualization
• Advanced recruiter analytics dashboard
• Company hiring metrics and ROI calculation
• Multi-language interview support
• Video interview analysis (body language, etc.)
• Industry-specific evaluation rubrics

📝 Project Structure

smartscreen/
├── README.md
├── .gitignore
├── backend/
│   ├── api/
│   ├── services/
│   ├── models/
│   └── config/
├── frontend/
│   ├── components/
│   ├── pages/
│   ├── utils/
│   └── styles/
├── ai-agents/
│   ├── task_generator.py
│   ├── technical_evaluator.py
│   ├── communication_evaluator.py
│   └── score_aggregator.py
└── docs/
    ├── architecture.md
    ├── api-spec.md
    └── deployment.md

🤝 Contributing

This is a college capstone project. Contributions welcome for:

• Bug fixes
• Feature suggestions
• Documentation improvements
• Code optimization

📧 Contact

Developer: Manu S
Project: SmartScreen - AI-Driven Screening Platform
Purpose: College Capstone + Hackathon Submission

---

Tagline: Fair, skill-based evaluation. No resumes. No cheating. Pure ability.

Where developers are judged on ability, not privilege.