CloudZenia Logo

How OpenSJLab Transformed Exam Integrity with AI Proctoring on AWS

Secure AI-powered assessment solutions for educational institutions, testing bodies, and EdTech providers.

Hero Background

Company:

Company: Logo

OpenSJLab is a leading AI-powered online live video exam and test monitoring platform creatively designed to ensure secure, transparent, as well as, scalable remote assessments. Their platform improves academic integrity by analysing candidates in real-time through video and audio streams. Besides, their platform also detects suspicious activities and generates detailed post-exam reports.

Industry

Financial Services (Banking)

Services

Amazon Bedrock, Anthropic Claude, RAG (Retrieval-Augmented Generation), Amazon ECS, AWS Fargate, Application Load Balancer (ALB), Amazon CloudFront, Amazon VPC, Amazon S3, Amazon RDS (PostgreSQL), Amazon ElastiCache, AWS Certificate Manager, AWS Secrets Manager, AWS IAM, SOC 2 Framework, Terraform, GitHub Actions, Pritunl

Share

facebook icon
x icon
linkedin icon
instagram icon

Business Requirements

OpenSJLab required a resilient, scalable, and exceptional infrastructure to support its swiftly growing AI-enabled exam monitoring solutions. The platform had to smoothly handle fluctuating workloads, particularly during peak examination windows. Also, concurrent test-takers should be able to expand to 10,000+ users, thereby assuring constant operation and availability alongside low-latency processing of live video and audio streams. Since achieving 99.9% uptime was a critical requirement, necessitating a highly available serverless architecture capable of automated adjustability and quick recovery during high-stakes assessments, was also utmost important.

Security and data integrity were paramount priorities, considering the critical nature of student data and the high stakes of academic certification. The infrastructure had to assure end-to-end encryption for all media streams, besides maintaining strict access controls for guaranteed privacy and also regulatory compliance. Furthermore, the platform required advanced Generative AI capabilities to move beyond simple motion detection, requiring a system that could provide contextual behavioural analysis and automated integrity reporting to reduce the manual burden on educational administrators.

To balance cost-efficiency with high performance, OpenSJLab needed a "pay-as-you-go" model that eliminated the overhead of idle servers during non-exam periods. The solution required intelligent resource allocation through AWS managed services and automated workflows to streamline the generation of post-exam audit reports. Additionally, the architecture demanded a safe and isolated data environment to ensure that critical exam artefacts were protected. Plus, they had to be completely accessible not only for institutional audits but also for global expansion across 15+ countries.

Image

The Challenge

Infrastructure Scalability and High Concurrency: Managing a large-scale EdTech platform requires the ability to handle enormous, unpredictable surges in traffic during global examination windows. The infrastructure needed to support 10,000+ concurrent video and audio streams without performance degradation. Switching to a serverless, event-driven model was fundamental for the system to scale dynamically while maintaining minimal lag for real-time monitoring.

Contextual AI Accuracy and Fraud Detection: Traditional monitoring tools often struggle with false positives, failing to distinguish between harmless environmental noise and actual misconduct. The challenge was to integrate Generative AI and advanced computer vision to deliver human-like contextual understanding, accurately detecting face absence, camera obstruction, and suspicious verbal activity across diverse lighting conditions and multiple languages.

Start Your Cloud Transformation Journey

CloudZenia supports you at every stage, delivering reliable, high-quality services at competitive prices.

The Solution

AI-Powered Monitoring through Amazon Bedrock and Rekognition

Generative AI was integrated via Amazon Bedrock to provide contextual analysis of candidate behaviour, moving beyond simple motion detection to understand the intent behind flagged activities.

Serverless Infrastructure and Expandability

To handle spikes of 10,000+ concurrent test-takers, the platform was built on AWS Lambda and Amazon ECS. This event-driven, serverless approach allows for spontaneous scaling during peak exam periods while eliminating the costs of maintaining idle servers. Moreover, Amazon CloudFront was utilised to deliver low-latency media streams to students across 15+ countries.

Automated Integrity Reporting and Language Processing

Amazon Transcribe and Amazon Translate were leveraged to capture and process multilingual audio, detecting unauthorized verbal communication. Generative AI models then synthesize these transcripts and visual flags into automated, concise Integrity Reports, reducing the manual review time for instructors by 60%.

Secure and Compliant Data Management

By utilising AWS Key Management Service (KMS) and AWS Secrets Manager, a robust security framework was implemented for encryption of critical student PII and exam artefacts. All data is securely stored in Amazon S3 with strict lifecycle policies and access controls, ensuring compliance with international education data privacy standards.

Identity and Access Control

Amazon Cognito was integrated to provide secure, scalable user authentication and authorisation. This ensures that only authorised proctors and students can access specific exam sessions, helping maintain a secure environment for high-stakes assessments.

Real-Time Observability and Performance Tracking

Amazon CloudWatch and AWS X-Ray provide extensive monitoring and tracing across the entire AI pipeline. This enables real-time detection of video processing congestion, ensuring consistent 99.9% uptime and an uninterrupted experience for users across global regions.

Result

Performance and Versatility Transforming into a serverless and event-driven architecture allowed OpenSJLab to expand swiftly, supporting 10,000+ synchronous test-takers without hampering performance. The integration of Generative AI through Amazon Bedrock simplified the analysis of exam sessions, reduced system latency by 40%, and assured real-time intervention capabilities during high-stakes assessments.
Improved Compliance and Security Implementing resilient encryption through AWS KMS and secure identity management via Amazon Cognito fostered trust with educational institutions. These stringent security measures ensured that sensitive student PII and exam artefacts were safeguarded, maintaining total compliance with global data privacy regulations and protecting against unauthorised access.
Operational Effectiveness By incorporating AI-driven automation and serverless workflows, OpenSJLab lowered manual invigilation effort by 60%. The automated generation of GenAI integrity reports accelerated the post-exam audit process, allowing educators to focus on academic outcomes rather than reviewing raw video footage, while enabling the platform to process over 1 million exam sessions annually.
Cost Optimization By utilising AWS Lambda’s pay-as-you-go model and optimising storage with Amazon S3 lifecycle policies, OpenSJLab significantly reduced cloud expenditure while maintaining elite performance. The elimination of idle server costs and the shift to a fully automated AI pipeline resulted in a 65% reduction in per-exam monitoring costs and a 35% optimisation in infrastructure spending, enabling 40% year-over-year revenue growth.
Results visualization showing target and metrics

Related services

Infrastructure as a Service

Infrastructure as a Service

Modern times need Modern Infrastructure Deployment

DevOps as a Service

DevOps as a Service

Embrace DevOps, Enhance Automation, and Deliver with Confidence.

Cloud Security

Cloud Security

Your Business is in the cloud? Secure it before its too late.

CI/CD

CI/CD

Your path to agile application deployment.

Cloud Native Development

Cloud Native Development

Simplify, Scale, and Save with Cloud-Native Application.

Automation

Automation

Unlock times treasure, labors burden lightened, by automations with errors few, and efficiency high.

Related Case Studies

LinkMe

LinkMe

Linkme is a social networking platform designed to facilitate easier and more effective networking by consolidating all of a user's social media profiles and links into one place LinkMe aims to simplify how people connect and share online...

Mormati

Mormati

Mormati, an emerging regional OTT platform, required a robust backend infrastructure capable of handling extensive media uploads, processing, and seamless content delivery to users on both mobile and web platforms...

Ready to Dive into Your Cloud Journey?

CloudZenia can help you wherever you are in your cloud journey. We deliver high quality services at very affordable prices.