Title
Core issue
Opportunities
Solution
Technical Approach
Feasibility and Viability
Thank you

ApplySure - AI-Powered Job Verification Platform

Inspiration

The genesis of this project came from a moment of realization that transformed concern into action. While watching the news coverage of Saudi Arabia's recent abolishment of the Kafala system, we witnessed the stories of millions whose lives had been upended. Workers who had traveled thousands of miles for promised opportunities were left only to find themselves trapped in cycles of exploitation and abuse.

This wasn't just a news story. It was a human crisis hidden in plain sight.

As we researched deeper, the scope of the problem became staggering. The Kafala system and many similar labor sponsorship structures across the Gulf, Southeast Asia, and beyond, had enabled decades of systemic worker exploitation. But the root of this exploitation often began far earlier: with fraudulent job postings, fake recruitment agencies, and manipulated contracts that lured vulnerable workers into situations from which there was no escape.

The Statistics Were Alarming

Millions of migrant workers from India, Bangladesh, the Philippines, Nepal, and Pakistan fall victim to fraudulent recruitment every year
Workers pay exorbitant fees (often $2,000-$10,000) for visas and placements that turn out to be fake
Upon arrival, passports are confiscated, contracts are switched, and workers find themselves in debt bondage with no legal recourse

This wasn't isolated to Saudi Arabia. Similar patterns emerged in Malaysia, Indonesia, Brunei, the UAE, Qatar, and Kuwait. The common thread was clear: fraud began at the point of recruitment, and once workers crossed borders, they became invisible.

The question that drove us was simple yet profound: What if we could verify the truth before workers left home? What if technology—the same AI that powers search engines and language models—could be redirected to protect the most vulnerable?

This wasn't just about building a verification platform. It was about restoring dignity, agency, and safety to millions who deserve to pursue opportunity without fear.

What it does

ApplySure is an AI-powered global job verification platform that authenticates overseas employment opportunities in real-time, protecting workers from fraudulent recruitment before they leave home.

Core Capabilities

1. Employer & Sponsor Verification

Validates company registration across 130+ jurisdictions using APIs from OpenCorporates, MOHRE, POEA, and government databases
Normalizes diverse employer identification systems (Kafeel IDs, UENs, EINs, POEA codes) into a unified verification layer
Generates Trust Scores (0-100) based on registration validity, complaint history, and cross-source consistency

2. AI-Powered Contract Analysis

Uses Google Document AI for OCR extraction from job offer photos or PDFs
Employs Vertex AI models to detect:
- Missing legal clauses (work hours, termination rights, salary guarantees)
- Salary mismatches against market averages
- Forged signatures or document tampering
- Inconsistencies in employer details

3. Multilingual WhatsApp Bot

Accessible via WhatsApp (no app download required) for low-literacy users
Supports 5 languages (English, Hindi, Arabic, Bengali, Tagalog)
Accepts voice messages and job offer photos
Returns instant verification results with actionable guidance

4. Real-Time Fraud Detection

Monitors emerging fraud patterns using machine learning
Alerts users, embassies, and government partners when suspicious agencies are detected
Builds a crowdsourced fraud database from verified worker reports

5. Multilingual Support & Translation

Integrated Google Translate API for automatic language detection and real-time translation across 5 languages
Uses Gemini LLM for natural language understanding and culturally-aware contract analysis
Ensures workers can interact with the platform in their native language while receiving contextually accurate verification results

How we built it

Our platform uses a microservices-based architecture on Google Cloud designed for scalability, real-time processing, and multi-stakeholder access across five integrated layers.

1. User Access Layer

Serves multiple stakeholders with tailored interfaces:

Job seekers upload documents via mobile or WhatsApp and receive real-time verification with chatbot-guided processes
Admins monitor system health, manage users, and configure verification rules
Recruitment agents submit bulk verifications and access analytics
Government partners retrieve fraud intelligence and compliance reports for integration with national e-migration systems

2. Microservices Layer

Implements three core services:

Employee Management handles authentication and encrypted profile storage with role-based access
App Services processes document uploads, manages sessions, and coordinates between frontend and backend
Agency Verification uses Polygon NET blockchain for secure identity verification with SHA256 hashing and maintains immutable audit trails

3. Backend Processing

Powers our fraud detection pipeline:

When users upload documents through OCR, AI analyzes contracts to identify suspicious patterns
Generates insights, flags high-risk documents, and provides actionable recommendations
Communication infrastructure spans Twilio API for WhatsApp Business, voice and text support for non-digital users, employer dashboards, chatbot interfaces, community forums, government call centers, and automated reporting that sends complaints to relevant authorities

4. Data & Intelligence Layer

Combines multiple database systems:

Vector Database stores embeddings for semantic search and fraud pattern matching
Relational Database maintains structured employer records and audit logs
RAG (Retrieval-Augmented Generation) combines real-time data retrieval with LLM reasoning
Gemini LLM processes natural language queries in 5+ languages and generates human-readable verification reports
Creates a local abstraction layer which provides unified query interfaces across all data sources

5. Integration & Security Layer

Ensures secure operations through:

SFTP connections for government data exchanges
Google Translate API for real-time translation
REST and GraphQL APIs for flexible data access
Confidence scoring that assigns verification results to graduated risk categories with color coding

Technical Stack

Category	Technologies
Frontend	React (web), Flutter (mobile), PWA
Backend	Django, PostgreSQL, MongoDB, Redis
DevOps	Docker, Kubernetes, Terraform, Cloud Build
Blockchain	Polygon network with smart contracts
ML/AI	Vertex AI, Gemini, TensorFlow, PyTorch
Cloud	Google Cloud Platform, Cloud Storage, BigQuery, Cloud Functions
APIs	Twilio, LiveKit, OpenCorporates, Google Maps

Challenges we ran into

Real-time Fraud Detection

Distinguishing legitimate small employers from sophisticated fraud without excessive false positives required a graduated risk scoring system: Verified (80-100%), Caution (40-79%), and High Risk (0-39%). We use user feedback loops to continuously improve model accuracy.

Blockchain Performance

Writing every verification to blockchain would create prohibitive costs and latency. Instead, we implemented selective anchoring where only critical results go to Polygon, intermediate processing stays in traditional databases, and non-urgent verifications batch processes run during off-peak hours.

Sensitive Data Security

Workers' personal information and whistleblower reports require extreme protection through end-to-end encryption, differential privacy in ML training, anonymized reporting, secure enclaves for document processing, and data residency compliance.

Multi-stakeholder Complexity

Building for workers, employers, agents, and governments with vastly different technical capabilities required modular access points. Low-literacy workers can use WhatsApp and voice interfaces while governments access analytics dashboards, all maintained through microservices architecture.

WhatsApp Bot User Experience

Users unfamiliar with chatbots can send unstructured queries or abandon conversations. We redesigned with guided yes/no flows, visual progress indicators, voice message support, automatic context retention, and fallback to human support at <70% AI confidence.

Data & Integration Challenges

Cross-border Data Access

Government databases are often closed or inconsistently formatted. Our hybrid strategy uses public APIs for baseline verification, simulation layers for demonstration, MoU templates for partnerships, and crowdsourced reporting as supplementary intelligence.

Language & Cultural Context

Employment contracts vary by cultural and legal context, not just translation. We fine-tuned Gemini on domain-specific datasets and implemented cultural context layers that flag clauses normal in one jurisdiction but exploitative in another, such as passport retention policies.

Accomplishments that we're proud of

🤖 Built a Fully Functional AI Contract Verification System – Successfully implemented Document AI for OCR text extraction and Vertex AI models that analyze employment contracts to detect missing legal clauses, salary mismatches, and suspicious patterns—the core intelligence powering our platform.

📱 Deployed a Working WhatsApp Bot Interface – Created and launched a functional chatbot using Twilio and Dialogflow CX that accepts job offer photos, processes them in real-time, and returns verification results in conversational language—proving our accessibility-first approach works.

🎯 Demonstrated End-to-End Verification Flow – Built a complete user journey from document upload through AI analysis to actionable results, validating that migrant workers can verify job offers in under 3 minutes using just their phones.

🏗 Designed Scalable System Architecture – Architected a microservices-based platform on Google Cloud with clear separation of concerns, positioning ApplySure for future expansion to government integrations and multi-country deployment.

🛡 Implemented Privacy-First Design – Built security into our foundation with encrypted data handling and anonymization pipelines, ensuring vulnerable users' information is protected from day one.

What we learned

Building for Vulnerable Users Changes Everything

We learned that creating technology for migrant workers requires fundamentally different design thinking. Our initial technical approach had to shift toward empathetic, accessible solutions that prioritize trust and clarity over sophisticated features.

AI Contract Analysis Is Harder Than Expected

Employment contracts have inconsistent formats, missing information, and ambiguous language. Training our models to handle real-world document variations while minimizing false positives taught us the importance of iterative testing and continuous model refinement.

WhatsApp Integration Unlocks Accessibility

Choosing WhatsApp as our primary interface wasn't just a technical decision—it was a philosophical shift. We learned that reaching vulnerable populations means meeting them on platforms they already use and trust, not asking them to adapt to us.

The MVP Reveals the Real Challenges

Building even a basic prototype exposed the complexity of cross-border verification: missing employer databases, inconsistent legal standards across countries, and the difficulty of providing actionable guidance without overwhelming users with information.

What's next for ApplySure

Enhanced AI Capabilities

Expand our fraud detection models to include image forensics for detecting photoshopped documents and implement multi-modal verification combining text analysis, metadata inspection, and visual authentication. Target: 95%+ accuracy with minimal false positives.

Offline Verification Module

Build offline-first capabilities allowing verification in areas with limited connectivity through cached employer databases and local processing, enabling community ambassadors to serve remote regions.

Extended Language Support

Add 10+ additional languages including Urdu, Nepali, Tamil, and Sinhala to reach underserved migrant communities across South and Southeast Asia.

Government Data Integration

Establish formal partnerships with labor ministries to integrate official employer registries via secure API connections, moving beyond our current public data sources to provide authoritative verification.

Worker Protection Features

Expand beyond pre-departure verification to include post-arrival support: wage tracking, grievance reporting mechanisms, emergency contact systems, and connections to legal aid networks.

Blockchain-Based Credentials

Launch ApplySure WorkerPass, a portable digital identity allowing workers to build verifiable employment histories that travel with them across borders and jobs.

Recruitment Agency Certification

Develop a verified agency program with public transparency dashboards, compliance training, and ethical recruiter ratings to incentivize good actors and isolate bad ones.

Immigration System Integration

Partner with visa-issuing authorities to embed ApplySure verification into official e-migration platforms, making fraud detection a standard step in the overseas employment process.

Global Employment Data Commons

Create an open, anonymized research dataset enabling evidence-based policy reforms, academic analysis of migration patterns, and predictive modeling of fraud hotspots.

AI Ethics Framework

Establish independent oversight with quarterly algorithmic bias audits, public transparency reports on verification accuracy, external security testing, and open-sourcing of non-sensitive components.

Our vision: A world where no worker's dream of a better life turns into a nightmare—where technology protects the vulnerable, not just the powerful.

ApplySure is more than a verification platform—it's a global movement toward ethical employment, transparent recruitment, and worker dignity. Every verification we process, every fraud we prevent, every life we protect brings us closer to that world.

This is why we built ApplySure. This is why it matters. This is just the beginning.