Sub-processors
The providers that help Ufuq operate.
Ufuq uses a small set of infrastructure and feature providers. This page identifies the categories of providers that may process data on our behalf and explains why each is used.
Last updated January 1, 2026
A sub-processor is a service provider that processes information for Ufuq while helping us provide the workspace. We select providers based on the function they perform, limit access to what is needed, and expect appropriate confidentiality and security commitments. The exact providers active for a deployment can depend on the environment and configuration chosen by the workspace operator.
| Provider category | Purpose | Data involved |
|---|---|---|
| PostgreSQL hosting | Database infrastructure | Account, workspace, application, and resume records stored by the deployment. |
| Hosting and delivery | Application hosting, static assets, and network delivery | Requests, IP addresses, logs, and encrypted application traffic. |
| Apify or the configured search provider | Search actor execution and result collection | Search criteria and source results when a user starts a search run. |
| Resend or the configured email provider | Verification and account email delivery | Email address, message metadata, and the verification or notification content required for delivery. |
| PostHog, when enabled | Product analytics and optional session insights | Configured product events and technical context. Analytics can be disabled by deployment configuration. |
| Cerebras, when Resume Studio generation is used | Server-side language model generation | The resume context and opportunity details needed to produce a requested draft; API credentials remain server-side. |
| Google Gemini, when internship match scoring is requested | Server-side text embeddings | Plain text extracted from the uploaded LaTeX resume and the saved internship titles, companies, and descriptions being ranked. Raw LaTeX commands and API credentials are not sent from the browser. |
| Tectonic compiler service, when compilation is used | Sandboxed LaTeX compilation | The LaTeX source and compile request needed to produce a PDF; the compiler is protected by a separate service key. |
Controls and changes
Search and resume providers receive information only when the corresponding feature is used. A database or hosting provider may process information as part of every authenticated request. Credentials for these services are held in server-only environment variables, and production transport is configured for HTTPS. The Tectonic service runs in a restricted container with a conservative LaTeX policy, no added Linux capabilities, resource limits, and a protected compile endpoint.
We may add, replace, or remove a provider as the product changes. When a change materially affects the categories of data processed, we update this page and the privacy policy. For a current data request, DPA question, or security concern, contact privacy@ufuq.app or security@ufuq.app.
Data minimization by feature
The workspace does not need every kind of information for every task. Account authentication uses identity and security information; opportunity organization uses the fields needed to describe a role; application tracking uses the notes and status a candidate chooses to enter. Resume generation, match scoring, and compilation are feature-specific actions, so their inputs are sent to the relevant provider only when the candidate requests that action. Match scoring strips LaTeX formatting and sends extracted resume text with saved internship titles, companies, and descriptions being ranked to the embedding provider. This separation helps keep an unused feature from receiving unrelated workspace data.
Providers are not allowed to use Ufuq information for an unrelated purpose through our service arrangement. We expect appropriate access controls, confidentiality, incident notification, and deletion or return procedures at the end of a service relationship. We review the operational need for each provider and avoid adding a vendor merely because it is convenient. If a provider changes for a deployment, the public list and relevant privacy notice are updated so account holders can understand the change.
The list describes processing categories rather than promising that every deployment uses every provider. A workspace operator can disable optional analytics or omit an integration that is not part of its configuration. The applicable privacy notice, account settings, and feature behavior are the best sources for understanding what is active for a particular workspace. Questions about a provider, transfer, or contract can be sent to the privacy address on our contact page.
A provider may receive information in a different country from the person using Ufuq. The privacy team considers the provider's role, the sensitivity of the information, the available contractual protections, and the technical controls around the connection. We do not treat a provider list as permission to send every workspace field everywhere; each feature should send only what its stated purpose requires.