Can I Self-Host an Accessibility Tool?

Yes, you can self-host an accessibility tool, and for many universities it is the right choice. Self-hosting gives your institution full control over document data, ensures compliance with privacy regulations like GDPR and FERPA, and keeps sensitive student records within your own infrastructure. However, self-hosting also comes with real operational costs that are worth understanding before you commit.

This post walks through why universities are increasingly interested in self-hosted accessibility tooling, what the landscape looks like today, and how to decide whether self-hosting fits your institution.

Why Universities Want Self-Hosting

The accessibility compliance space has been dominated by cloud-based SaaS products for years. So why are IT teams at universities pushing back?

Data Sovereignty and Student Privacy

Universities handle some of the most sensitive personal data in any sector. Course materials may contain student names, grades, accommodation records, and health information. When you upload a PDF to a cloud remediation service, that data leaves your network and enters someone else's infrastructure.

For institutions bound by FERPA in the United States, this creates a compliance question: does uploading student-facing documents to a third-party processor constitute an unauthorized disclosure of education records? The answer depends on your vendor agreements, but many compliance officers prefer to avoid the question entirely by keeping data on premises.

In the European Union and Australia, GDPR and equivalent frameworks impose strict requirements on cross-border data transfers. A university in Sydney sending documents to a US-based accessibility platform may need to navigate Standard Contractual Clauses, data processing agreements, and transfer impact assessments. Self-hosting eliminates that friction.

Institutional Control

Universities operate on long timescales. A cloud vendor might change pricing, deprecate features, or shut down entirely. When your accessibility infrastructure runs on your own servers, you control the upgrade timeline, the data retention policy, and the integration points with your LMS and document management systems.

What Self-Hosting Actually Means

Self-hosting an accessibility tool means running the software on infrastructure you control, whether that is a physical server in your campus data centre, a virtual machine on a cloud provider you manage, or a container orchestration platform like Kubernetes.

In practice, a self-hosted accessibility stack typically includes several components:

• A document processing engine that can parse PDFs, Word documents, PowerPoint files, and spreadsheets
• An OCR layer for scanned documents and images
• A validation engine that checks documents against WCAG 2.1 standards
• An AI model for generating alt text, fixing reading order, and remediating tagged PDFs
• A web interface for staff to upload documents and review results

Each of these components needs to be installed, configured, secured, and maintained.

Open-Source Tools You Can Use Today

Several mature open-source projects cover parts of the accessibility pipeline:

veraPDF is the industry-standard open-source PDF/UA validator. It can check whether a PDF conforms to accessibility standards and produce detailed reports. It is well-maintained and widely trusted, but it only validates. It does not fix problems.

Tesseract OCR is the leading open-source optical character recognition engine. It handles text extraction from scanned documents and images across dozens of languages. It is a critical building block, but it requires integration work to fit into a document remediation workflow.

LibreOffice can convert between document formats on the command line, which is useful for processing Word and PowerPoint files. Combined with Python libraries like python-pptx and openpyxl, you can build format-specific processing pipelines.

The challenge is that these tools are components, not products. Assembling them into a coherent, reliable accessibility remediation system is a significant engineering project.

The Challenges of Self-Hosting

Self-hosting is not free, even when the software is open source. The real costs show up in three areas.

Maintenance and Updates

Accessibility standards evolve. WCAG 2.2 added new success criteria. PDF/UA-2 is gaining adoption. Your self-hosted tooling needs to keep pace, which means someone on your team needs to track standards changes, update validation rules, and test against new document types.

AI Model Hosting

Modern accessibility remediation relies heavily on AI for tasks like generating alt text for images, inferring reading order in complex layouts, and suggesting table header structures. Running these models locally requires GPU hardware or significant CPU resources, along with expertise in model deployment and optimization. For a deeper look at what AI-powered remediation involves, see our guide to choosing a PDF remediation tool.

Integration and Support

Your accessibility tool needs to work with your LMS (Canvas, Blackboard, Moodle), your cloud storage (Google Drive, OneDrive, SharePoint), and your institutional authentication system. Each integration is a project in itself, and each needs ongoing maintenance as upstream APIs change.

Cloud vs Self-Hosted: The Tradeoffs

When choosing between cloud and self-hosted, consider these factors:

• Data location: Cloud puts data on vendor infrastructure; self-hosted keeps it on yours
• Privacy compliance: Cloud depends on vendor agreements; self-hosted is fully within your control
• Setup time: Cloud takes minutes; self-hosted takes days to weeks
• Maintenance burden: Cloud is vendor-managed; self-hosted requires your team
• AI capabilities: Cloud uses vendor-managed models; self-hosted means you host and update models
• Cost model: Cloud charges per-document or subscription; self-hosted costs infrastructure and staff time
• Customisation: Cloud offers limited vendor options; self-hosted gives full control

Neither option is universally better. The right choice depends on your institution's privacy requirements, technical capacity, and budget.

Infrastructure Requirements for Self-Hosting

If you decide to self-host, plan for the following minimum infrastructure:

• Compute: A server with at least 8 CPU cores and 32 GB RAM for document processing. Add a GPU (NVIDIA with CUDA support) if you want to run AI models locally.
• Storage: Fast SSD storage for temporary document processing, plus archival storage for scan history and audit logs.
• Networking: HTTPS with valid certificates, integration with your institutional identity provider (SAML or OIDC), and network access to your LMS APIs.
• Containerisation: Docker or Podman for running isolated processing components. A reverse proxy like Traefik or Nginx for routing.

For detailed specifications, see our self-hosting documentation.

Aelira's Approach

Aelira was designed from the start to run on your infrastructure. The platform deploys as a set of Docker containers behind a reverse proxy, processing documents locally without sending data to external services. AI models run on your hardware, and all document data stays within your network.

This means you get the convenience of an integrated accessibility platform, covering PDF remediation, Office documents, LaTeX, images, and more, without compromising on data sovereignty. Updates ship as container images that you pull and deploy on your own schedule.

For institutions that want the benefits of self-hosting without building a custom toolchain from scratch, this approach offers a practical middle path.

---

If your university is evaluating self-hosted accessibility solutions, get in touch with the Aelira team to discuss how the platform can fit your infrastructure and compliance requirements.