Privacy Policy | Job Pocket

Job Pocket helps you save jobs you see on job sites or in Discover, summarizes the key points of each job, evaluates whether it matches your job-search preferences, and helps you write cover letters, prepare for interviews and practice with a voice mock interview. This policy explains what data the app handles, where it is stored, which service providers it is handed to for processing, and how you can delete your data.

You can use the basic saving features as a guest. The app creates a random installation identifier, obtains an access credential from the server, and stores the credential in the iOS Keychain. If you choose to sign in with Apple, the server stores the user identifier provided by Apple and, where Apple provides it, your email address; guest usage is also merged into the account to avoid counting quotas twice. Pro subscriptions are paid through Apple; we store the product identifier, the transaction association, the validity period and the revocation status in order to verify your entitlement, and we do not receive your credit card details.

Data stored on your device

The jobs, resumes and personal settings you save are stored mainly on your iPhone and use iOS data protection. When you use AI-related features, only the necessary data described in the sections below is sent:

Job links, job content and AI analysis results you have saved (including jobs saved from Discover)
Job-search preferences (job categories, work locations, expected salary, nickname)
The target position, years of experience, skills, work history and education that you confirm or enter manually in My Resume; the original imported text is not kept
Application progress and in-app notifications
Cover letter content and writing settings, and the interview key points for each job
Interview plans, transcripts, interview reports, and private interview notes you enter yourself
App settings such as appearance (light/dark)

When you send a job to Job Pocket from the share sheet, the URL, page title and page text are first held temporarily in an App Group space that only Job Pocket can access. After you open the app, they are moved into the local data above and cleared from the temporary space.

Data stored on the server

The server stores the account identification data needed to provide the service and prevent abuse, the creation and last sign-in times, subscription entitlements, and the monthly usage counts for job analysis, cover letters and mock interviews (without content). Interview quota records include a session identifier that does not contain the transcript and the number of report attempts; report access expires 7 days after an interview starts, with at most 3 attempts per interview. We also store service-wide daily usage that contains no personal content, for up to about 7 days. Feature requests, likes, reports and blocking data are also stored on the server. During an ongoing voice mock interview, the conversation text is held temporarily in server memory so that the interviewer can continue asking questions; it is cleared when the interview ends, for no longer than about 3 hours, and is not written to a database.

To speed things up and reduce repeated reads, the server caches, per job URL, publicly readable job content (job title, company, job description, requirements and so on) from the 104 Job Bank, LinkedIn, Cake, Yourator and 1111 websites (104 Job Bank, Cake, Yourator and 1111 are Taiwanese job sites), for at most 14 days. This cache contains only public job information and does not record who saved or analyzed the job. The page text you share, your job-search preferences, your resume, your personal fit analysis results, your cover letters and your interview key points are not written to the server database.

To limit abnormally large volumes of requests, the server briefly uses the connecting IP address in memory to calculate request frequency; it is not written to a database. Operations systems may store technical information such as request IDs, API paths, status codes, source platform, response time and usage, and may notify operators through a private Telegram notification; resumes, original job text, access credentials and API keys are not attached. Our network service providers (for example Cloudflare and hosting providers) may record connection information under their own policies to maintain security.

Data sent for AI processing

After you explicitly consent to having the necessary data processed by AI services, when you use the following features the app sends content over HTTPS to the Job Pocket server, which then hands it to an AI provider for processing:

Job analysis (analyzing a job or requesting re-analysis): the job URL and job content (the page text you shared, public job data the app read on your phone, or the content of a Discover job), plus, where you have filled them in, your job categories of interest and target position, skills, years of experience, preferred work locations, work arrangement, job type, expected salary and resume content.
Cover letter: the job content, your resume content, current position, years of experience and skills, the signature you entered, the language (Traditional Chinese or English), tone and length settings, and the points you want to emphasize; when revising, also the previous version and your revision instructions.
Interview key points: the job content, the public company data the app already has, and your resume content.

When you save from the share sheet, the share screen only reads public job information and stores your choice; it does not run AI analysis. If you choose “Analyze”, it starts when you open the main app, according to your consent status. If you choose “Save link only”, it may read the public page, but it does not send your personal resume or preferences to AI.

The fit score is calculated by the server by comparing job requirements with your job-search preferences according to rules; the server then hands the job key points, the comparison results and the data above to an AI provider to organize the information and generate text. Currently OpenAI is the primary model, and when it is unavailable processing falls to Google Gemini; the actual provider depends on service settings and availability. When OpenAI is used, response storage is turned off; this does not mean the provider retains no data at all. Google Gemini is used through a paid API project with billing enabled; under its paid-service terms, inputs and responses are not used to improve Google products, but the provider may still keep records for a limited period for security and abuse prevention. The AI-generated fit scores, cover letters, interview key points and suggestions are for reference only; please decide for yourself whether to apply, and check that a cover letter is accurate before using it.

We use API services that have applicable data processing terms and security measures, and we require third parties to give the data they receive the same or equivalent protection as this policy; we do not authorize providers to use the data above for advertising or to sell it. Google’s processing is based on the Gemini API Paid Services Terms and its data processing addendum. OpenAI API data is not used to train models by default; the default retention period for security and abuse-monitoring logs is generally up to 30 days, with legal or security exceptions handled under its API data controls policy. A provider’s security logs are different from Job Pocket’s application database, and we cannot promise zero retention.

Please do not put unnecessary sensitive data, such as national ID numbers, passwords or bank account numbers, in job content or your resume.

Reading job URLs

When you save a job from 104 Job Bank, the app first opens that public job page on your phone in a built-in browser that does not keep cookies, and reads the job data on the page; if reading fails, the Job Pocket server reads the same public URL through Jina AI Reader (r.jina.ai). Public job pages on LinkedIn and Cake are read directly by the Job Pocket server, and Jina AI Reader may also be used when reading is blocked. For Yourator and 1111, the public job is read first, and if the page has not yet rendered its content, it may be rendered through Jina AI Reader; if a login or security check is encountered, reading stops. Greenhouse job content is read through its official public Job Board API. For Glassdoor, only the link you saved is kept and the job text you actively share or paste is processed; it is not fetched by the server and is not written to the shared job cache. Job links outside the allow-list are not accepted. These reads use only the job URL and do not send your preferences, resume or other personal data.

Discover jobs and company data

The jobs shown in Discover are compiled by Job Pocket from the public job pages of about 80 companies on Greenhouse, Lever and Ashby, keeping only jobs posted within the last 14 days; company logos are taken from each company’s website. Company founding year, headquarters, size, industry and listing information are compiled from public data on Wikidata and Wikipedia and may be incomplete or out of date. When you browse Discover and view company data, the app obtains this compiled public content from the Job Pocket server, and does not send your resume or preferences to these job platforms or to Wikidata or Wikipedia. Tapping “Save” saves the job only to your phone; it is sent as described in the previous section only when you choose to analyze it. Applying takes place on the employer’s original job page, and that website handles your data under its own policy.

Resume auto-organizing from an upload

When you use “One-tap resume organizing”, the app first converts the PDF or plain-text file into text on your phone (for a PDF, the embedded text is read directly); the file itself is not uploaded. Only that text is then sent over HTTPS to the Job Pocket server, where an AI service organizes it into a target position, current position, years of experience, skills, work history and education, and returns the result to the app immediately. The server does not store the resume text or the organized result; the AI is instructed to skip personal data such as contact details, national ID numbers, addresses and birth dates. The organized result is first shown to you for confirmation, and only after you confirm and save is it written to the career fields on this phone; the original imported text is not kept. You can also fill in or edit the fields manually; saving manually does not call AI.

Voice mock interviews and interview intelligence

Before a mock interview starts, your explicit consent and microphone permission are requested separately. The interview is turn-based: each time you answer a question, the app records that answer on your phone and processes it over HTTPS through the Job Pocket server, as follows:

Speech to text: the recording of your answer is handed to Groq’s speech recognition service (Whisper-family models) to be converted to text; when Groq is unavailable, OpenAI’s speech recognition is used instead.
The interviewer’s questions and replies: the transcribed text, together with the job content, your career data (target position, years of experience, skills, resume content), the interview key points for that job and the interview instructions, is handed to a language model hosted by Groq to generate the interviewer’s next line; the questions are also drafted by this model based on the job and the resume. When Groq is unavailable, OpenAI and then Google Gemini are used, in that order.
The interviewer’s voice: the text the interviewer is to speak is handed to ElevenLabs to synthesize speech; when it cannot be synthesized, Google Gemini synthesizes it, and when neither is available the Apple voice on your phone reads it aloud. Only the interviewer’s text is sent for speech synthesis, not your recording.
Live captions: the live captions shown on screen while you speak use Apple’s speech recognition, which requires separate “Speech Recognition” permission. When the device supports it, recognition is done on the phone; when it does not, Apple may process the speech on its servers under its privacy policy. Without this permission you can still record answers, but live captions are not shown.
Interview report: after the interview ends, the transcript and the questions are handed through the server to OpenAI to generate a written report and score; when it is unavailable, Google Gemini is used.

The Job Pocket server does not store the original audio and does not write transcripts or reports to a server database; the conversation text in progress is held in server memory only during the interview. Transcripts and reports are stored on your phone. Groq, ElevenLabs, OpenAI, Google and Apple may keep data for a limited period for security and abuse prevention under their own policies.

Interview intelligence consists of public job information, your private interview notes, and search links that lead to third-party websites. Company data may come from the public company pages of 104 and is read through Jina Reader; only the public URL is sent, not your resume. After you tap a third-party search link, that website handles the search terms and connection data under its own policy. AI content is for interview preparation reference only and does not represent an interview process confirmed by the company.

Public feature requests, reports and blocking

When you submit a feature request in Notifications → Community & Updates, the title, content, category and the display name you choose are sent to the server and are made public immediately to other users, without prior review; we act when we receive a report or discover a violation. Pornography, threats of violence, hate, harassment, fraud, spam and content that violates others’ privacy are prohibited; violating content is removed, and serious or repeat violators are suspended. Please do not post personal contact details, resumes or sensitive data about third parties. Your internal account identifier is not made public. Likes, reports and blocks are associated with your account and are used to remove duplicate actions, hide content you do not want to see and handle abuse. You can delete your own feature requests, or report and block the author from the content menu; you can also contact us through the support email below.

What we do not do

Job Pocket contains no ads, uses no cross-app tracking, does not sell or rent personal data, and does not provide your data to job sites or companies. The AI cannot access your contacts, calendar, location or arbitrary device files; the microphone is used only when you answer in a mock interview.

Retention and deletion

Device data is kept until you delete it. You can delete a resume, a job or interview records individually, or delete your account in Profile → Settings → Delete Account and All Data. Users who signed in with Apple must verify their identity again to revoke Apple authorization; only after the server has successfully deleted the account, the related guest data, usage, subscription associations, feature requests, likes and report identification records does the app clear the local data and sign-in credentials. If deletion fails, the app shows an error and keeps the local data so that you can retry. Deleting the account does not automatically cancel your Apple subscription; please cancel it separately in Apple’s subscription management.

Please use the in-app deletion feature first; simply removing the app does not guarantee that the Keychain credential or the server account is cleared. Monthly usage records are set to expire and be purged after about 1 to 2 years; report access expires after 7 days, but its non-content usage records are still kept for this period. Account, subscription association and community data are kept until you ask for deletion or until they are no longer needed to operate the service. The public job cache, which is not linked to accounts, is kept for at most 14 days. If you cannot use in-app deletion, contact the email below; we will respond within 30 days and handle verifiable requests.

Deletion cannot withdraw processing requests already sent to AI providers; security logs that providers keep under their policies, and Apple’s transaction data, are also not managed directly by Job Pocket. In the privacy page in Settings you can withdraw your AI consent, which stops later AI analysis, cover letters, interview key points and voice interviews.

Children’s privacy

Job Pocket is intended for adult job seekers; AI analysis, resume organizing, cover letters, interview key points and voice mock interviews are only for users aged 18 or older, and before use you must explicitly confirm your age and your consent to data transmission. We do not knowingly collect personal data from children below the legal age of consent in their location.

Policy updates and contact

If there is a material change in how data is handled, we will update the date on this page and, where appropriate, notify you in the app. If you have any questions about privacy or data deletion, email [email protected].

Guest use, Sign in with Apple and subscriptions