When we first met, Rosalyn was an edtech startup with proven AI and computer vision capabilities and a mission to keep assessment unbiased. They needed a tech partner who’s skilled in product design and development and knows how to tackle the pain points of early-stage ventures. Having examined our portfolio, skills, and testimonials, Rosalyn entrusted Merixstudio with the following tasks:
We’ve been working on the MVP since December 2019. The usefulness of Rosalyn’s concept has already been recognized, as our client raised seed round funding totaling $2.2MM from private investors in November 2020.
seed round funding
months of collaboration
designs delivered
team members contributing to the project
Our partnership started with a two-day workshop, during which Merixstudio’s team of 3 developers, a UX designer, and a PM talked to the client’s representatives – CEO, CTO, and the Product Manager – to uncover their needs and expectations. Sorting out priorities allowed us to proceed with wireframes and designs, which laid the foundations for the smooth development.
Next, we put together a cross-functional team whose task was to build a complex yet scalable system. Cooperating with the client’s AI team, our developers have created an online live proctoring app and come up with multiple fraud prevention strategies within the desktop app, such as prohibiting screenshots, using external apps throughout the exam, and forcing users to unplug the external monitor before starting the exam.
The desktop application makes an essential part of Rosalyn’s platform, as it supports different operating systems and facilitates live streaming. It also sustains the exam session even in case of temporary connection breakdown, which comes in handy considering different locations of the examinees. The app is Electron-based, which translates into a wider developer talent pool accessible for development and future maintenance.
Ever since the onset of the project, security has been one of the main concerns. On the one hand, the client established strict access permission rules and continues to educate our team on cybersecurity threats and best practices. On the other hand, Merixstudio’s QA specialists followed rigid rules concerning user data protection.
Among the most prominent project challenges, there’ve been load tests, which – due to their complexity – required the creation of a dedicated tool. Another challenge was integration with Vonage Video API (formerly known as TokBox), which enables video sharing between the user, AI-powered app, and proctoring panel. Last but not least, as is often the case in the startup world, now and then, project requirements change – Merixstudio’s team manages these changes thanks to Agile development.
Now, we’re focusing on delivering an open API and utilizing the LMS LTI standard to enable the integration of Rosalyn’s tool with external learning management platforms such as Canvas, Moodle, or Brightspace. This calls for two things: making the platform fully LTI compliant in terms of user interaction with Rosalyn’s system on the one hand, and ensuring that online proctoring within the app takes place regardless of the test engine on the other hand.
To achieve the best results possible, we stay flexible in terms of the adopted workflow. Merixstudio’s development team works hand in hand with Rosalyn’s AI team supervised by the client’s tech leads. Simultaneously, we don’t refrain from suggesting improvements related to the adopted tech solutions and time and effort management. We also proactively consult with Rosalyn’s Product Managers about requirements for features yet to come. As far as our daily job is concerned, we follow Scrum, organize our work into two-week sprints, and use tools like Jira or Slack to ensure full transparency.
for creating and managing examinations in the authoring panel, the app currently uses an external assessment provider. The other two authoring-related steps involve creating the exam groups and the complete assessments that consist of several exams
the learner portal allows students to examine the assessments that have been assigned to them. This way, they can track their progress, preview the scoring, and see the proctoring results
learners take exams in the desktop delivery app, which forces them to follow specific exam rules. On the one hand, it prevents students from using certain apps, browsing through the Internet, or copying/pasting content during the exam. On the other hand, the app is connected to an AI-powered proctoring tool that can perform live and post-processed analysis
aI monitors the examinee’s interactions, as well as computer video and audio to ensure test integrity while taking the test. Any suspicious actions, such as talking to other people or using cheat sheets, external screens, and many more, are flagged and subject to further proctoring inspection
the app has a built-in proctoring portal that professional proctors can use to inspect the proctoring events and decide to pass or fail users’ exams due to the reported violations. This includes both live proctoring mode, where the human and AI inspections are performed simultaneously in real-time, and post-processing mode, where the analysis is conducted after the exam. Live proctoring features also enable pausing and stopping the exam and sending warnings to learners during the exam attempt
.svg)
