Automate
A safety-first job application workflow
Automate receives job posts from Telegram, extracts and validates opportunities, records durable state, and prepares truthful application materials. An explicit Telegram approval is required before the system sends an email application.
Context
The problem space.
Job opportunities arrive in inconsistent formats and repeatedly require extraction, evidence checking, document preparation and a controlled sending decision.
Confirmed work
What the evidence supports.
- A Telegram intake workflow handles channel posts and approval callbacks, including posts that contain more than one opportunity.
- Structured LLM extraction is bounded by deterministic validation, correction limits and stable PostgreSQL application records.
- The system generates and validates truthful LaTeX resume artifacts before an application package can be approved.
- A Telegram approve/reject step gates Gmail delivery, while Google Docs provides the human-facing tracker.
System flow
- 01
Telegram job posts
- 02
Multi-job extraction & validation
- 03
PostgreSQL application state
- 04
Truthful resume tailoring & PDF validation
- 05
Telegram approval
- 06
Gmail application send
- 07
Google Docs tracking
Proposed / planned
The direction, not a completion claim.
- Embedding-based retrieval and Google Drive artifact upload remain explicit future extensions.
- Cloud Run inference requires separately configured cloud credentials and controlled live verification.
Evidence note
Repository evidence documents local validation and inactive workflow imports. Telegram, Gmail, Google Docs and Cloud Run integrations still require configured credentials and controlled live verification.