Home > Solutions > PANS Jarosław – multimedia materials repository for university

PANS Jarosław

Web-based multimedia repository platform for a public university – featuring video and audio streaming, automated AI-generated subtitles, and integration with USOS and SSO, all hosted entirely on-premises.

About the project

For a public university, we designed and implemented a web-based multimedia repository platform that enables the publishing, searching, and playback of video, audio, and digital document materials.

The system was built as a digitally accessible platform meeting WCAG 2.1 requirements, with a responsive interface that works in popular browsers and support for teaching materials, lectures, and event recordings.

As part of the project, we carried out a pre-implementation analysis, architecture design, and complete system configuration and parameterization, including integration with the USOS system based on views in an Oracle database. The software supports advanced permission management (public access, login, CAS SSO, custom accounts), automatic user account creation, time-based access restrictions, and mechanisms for assigning rights based on USOS data (field of study, faculty, class group).

Business goal

The goal of the project was to build a central system that standardizes the sharing of multimedia materials with students and university staff in a manner compliant with digital accessibility requirements (WCAG 2.1) and GDPR.

The university needed a system that would:

– meet digital accessibility requirements (the Digital Accessibility Act and WCAG 2.1) – as a public university, it is required to provide recordings with subtitles,

– enable the centralization of dispersed materials – until now, lecturers had been using various tools and drives, without a common standard,

– ensure data security – access control, image/likeness protection, GDPR,

– support the university’s digitalization – the project is part of a broader IT infrastructure modernization program.

The system was built from scratch in a new area.

Design challenges

The project required combining several areas in which close cooperation with the university’s infrastructure and team was essential.

Integration with the USOS system was implemented through dedicated views in an Oracle database. Cooperation with the university’s IT department involved determining the limited scope of data needed in the application and a secure method of transferring it, so as to synchronize students, staff, and class groups while keeping the amount of data to a minimum.

A key assumption was full data locality. A proprietary media server and the AI transcription component (Whisper) were deployed within the university’s infrastructure, so that no materials would leave the server room or be sent to external clouds. This required preparing the entire process – from upload, through transcoding and subtitle generation, to availability in the player – as a fully automated and asynchronous workflow running in the background.

Security was a critical concern. The digital accessibility requirements of WCAG 2.1 and GDPR compliance were addressed on an ongoing basis throughout the implementation, rather than merely audited at the end, which included, among other things, implementing functionality supporting the “right to be forgotten.”

An organizational challenge was also the physical deployment of the hardware environment on site – installing the server in the university’s existing rack cabinet and connecting it to the network, carried out in cooperation with the university’s IT department.

Our approach – scope of the solution

The platform was designed as a single, cohesive environment for publishing, managing, and playing back audio-video content and digital documents.
Among other things, the project delivered:

  • Advanced search
    • filtering by material type, categories, keywords, and publication dates
    • quick preview of content without logging in
  • Streaming and playback
    • video and audio streaming (HLS, HTML5) with adaptive quality (Full HD / HD / SD)
    • playlist support
    • works in every browser, on desktop and mobile
  • Permission management
    • various access models: public, authorized, CAS SSO, custom accounti
    • automatic user account creation
    • time-based access restrictions
    • assigning rights based on USOS data (field of study, faculty, class group)
  • Automatic AI transcription
    • automatic generation of Polish-language subtitles for audio-video materials
    • Whisper (AI) engine running 100% locally on the university’s server — without sending recordings to external clouds
    • asynchronous processing, completion notifications, and the ability to link .srt/.vtt files to video material
  • Admin panel
    • comprehensive management of materials, users, and permissions
  • Process automation
    • the entire process is fully automated and asynchronous
    • a lecturer uploads material and receives an email notification when it is ready
    • preparation and encoding run independently and in the background

Architecture and technologies

Technology stack
  • Backend: NestJS
  • Frontend: Next.js
  • Baza danych: PostgreSQL
  • Streaming: HLS (NestJS / Nginx)
  • Environment: Docker
  • AI components: Whisper (run locally on the university’s server)
  • Integrations: USOS, CAS (SSO)

This is a proprietary solution built from scratch, integrating with open-source libraries such as FFmpeg (preparing files for HLS streaming) and Whisper (the Speech-To-Text process)

System architecture

The system operates in a modular architecture: a central backend works together with separate, independent services for transcoding and transcription. This approach is easy to maintain and scale.

A proprietary media server was deployed within the university’s infrastructure – no materials leave the server room. HLS streaming is delivered with adaptive quality (Full HD / HD / SD).

Integration with USOS takes place through dedicated Oracle views – students, staff, and class groups are synchronized. Cooperation with the university’s IT department involved determining the limited scope of data needed in the application and a secure method of transferring it. Login is handled through the university’s SSO (CAS) – integrated via OIDC, so that users rely on the same credentials as in the university’s other systems.

The architecture was prepared for expansion – adding further integrations, growth in the number of users and materials, additional streaming nodes, and integration with new university systems at a later date.

Performance and safety

The system is hosted on a server located in the university’s server room. Full data locality means that materials are kept entirely on the client’s server.

The security mechanisms implemented include:

  • roles and access groups, including access aligned with permissions granted based on the USOS integration,
  • a secure authentication mechanism (Cookie-Based Authentication),
  • HTTPS encryption,
  • the performance of security and vulnerability testing, together with the delivery of a report in this regard,
  • anonymization of users in accordance with GDPR, including the implementation of functionality supporting the “right to be forgotten.”

The WCAG 2.1 digital accessibility requirements were addressed on an ongoing basis throughout the implementation. At the final stage, we conducted digital accessibility testing and prepared an accessibility statement published on the website, along with a test report delivered to the client in document form.

Implementation results
  • centralization of sharing multimedia materials with students and university staff
  • compliance with digital accessibility requirements (WCAG 2.1) and GDPR
  • automatic generation of Polish-language subtitles using AI, running fully locally
  • integration with the USOS system and the university's SSO (CAS)
  • full data locality — materials do not leave the university's server room

Więcej projektów

Porozmawiajmy o Twoim projekcie

Opisz swoje potrzeby, a my wrócimy z propozycją dopasowaną do Twojego biznesu.

Contact