Your CV stays in your browser

Plain-language summary

• AI inference and document parsing run in your browser.
• No account is required.
• The app does not set advertising or analytics cookies.
• Optional workspace saving is off by default and stores data only in this browser’s private site storage.
• Libraries, fonts, and AI model files come from external hosts, which may receive ordinary request metadata such as your IP address.

Local processing

PDF, DOCX, TXT, and pasted text are read by JavaScript running on your device. The selected language model runs locally through WebGPU. The application does not contain code that posts your source text or generated draft to a remote AI service.

You remain responsible for deciding what information to enter. Avoid unnecessary confidential, personal, Indigenous, community, partner, trainee, or third-party information.

Browser storage

WebLLM caches downloaded model files in browser-managed storage so later visits do not always require another multi-gigabyte download. Your browser controls this cache.

The optional “Save this workspace in this browser” setting stores CV text, verified evidence, selections, and drafts in IndexedDB on this device. It is opt-in and disabled by default. Use “Clear saved browser data” in Settings, or clear this site’s data in your browser, to remove it.

Network requests and third parties

The executable application, WebLLM runtime, PDF reader, and Word reader are pinned and self-hosted on this website; the page does not fetch executable JavaScript from jsDelivr or esm.sh at runtime. Fonts and model files may still be downloaded from providers including Google Fonts and model-hosting infrastructure such as Hugging Face. Those providers may log standard technical information under their own privacy policies. “Local AI” means your prompt and CV content are processed locally; it does not mean the first visit makes no network requests.

After the first successful load, a service worker caches the public application shell for offline use. It does not cache your CV, extracted evidence, drafts, or optional IndexedDB workspace. WebLLM manages model-weight caching separately.

Desktop option

Local TCV Desktop runs a Python API on 127.0.0.1 only. CV text used for inference is sent to that localhost process (and onward only to local Ollama or llama-cpp), not to this website’s server or a cloud model API. Model pulls through Ollama or GGUF hosts may still involve ordinary download metadata.

Server logs

The website host may keep ordinary access and security logs, such as requested URLs, time, IP address, browser information, and error details. These logs should not contain your CV text because the app does not upload it.

Your choices

You can use guided mode without uploading a file, paste only selected material, leave optional browser saving off, clear saved data, and close the page at any time. Clearing browser site data may also remove downloaded model files.

Questions can be directed to Sidney Shapiro.

Last updated July 25, 2026.