How we protect your data.
NDAs before any data moves, need-to-know access, and the option to keep everything inside your own tools.
AI teams trust us with data that is often unreleased, proprietary or sensitive. These are the practices we follow on every engagement. Where a client needs more, we agree project-specific controls in writing.
Confidentiality first
- We sign an NDA or confidentiality agreement before any data is shared, including for free pilots.
- Everyone on our team works under a confidentiality agreement.
- We never use client data to train our own models, sell it, or share it with other clients.
Access control
- Access to project data is limited to the annotators, reviewers and QA leads assigned to that project.
- Access is removed when a team member leaves the project or the engagement ends.
- Team members are trained not to copy, download, screenshot or share project data outside the approved tools.
Work in your environment
We're tool-agnostic. Whenever you prefer, we work directly inside your own labelling platform (CVAT, Labelbox, Encord-style platforms or in-house tools), so your data stays in systems you control and you decide who can see it.
Data handling
- Data is moved only through secure, encrypted channels you approve.
- We use client data only to perform the work you've asked for.
- At the end of a project, we return or delete any client data we hold, and confirm deletion on request.
Personal and sensitive data
For projects that involve personal data, such as speech, faces or documents, we act on your instructions and follow Kenya's Data Protection Act, 2019. We can work with your data processing agreement and any extra controls your project needs.
Questions or a security review
If you have a security questionnaire, need a data processing agreement, or want to talk through your requirements before a pilot, email joseph.mwanzia@saarve-ai.com.