Continuous Job Intake Assistant

This repository now focuses on continuous job-intake automation with a required human review step. It can:

1. Tailor a truthful resume draft from a job description.
2. Export both an ATS-friendly PDF and a LaTeX source file.
3. Poll an Outlook inbox through Microsoft Graph for Handshake or employer application links.
4. Prepare per-job application packets for review instead of blind auto-submitting.

Why the review step matters

The automation in this repo is intentionally designed to prepare materials for your approval, not submit applications without you seeing them. That keeps the workflow aligned with truthful self-representation and gives you a chance to verify compensation, location, work authorization, and any screening questions before submitting.

Project layout

• index.html, app.js, styles.css: browser UI for one-off tailoring.
• generate.py: simple HTTP handler for the frontend.
• resume_service.py: resume tailoring, LaTeX generation, and JSON helpers.
• pdf_renderer.py: lightweight ATS-friendly PDF rendering.
• outlook_watcher.py: Outlook inbox polling plus reviewed application packet generation.
• main.py: CLI entry point for one-off tailoring or continuous inbox polling.

Outlook inbox polling

The watcher uses the Microsoft Graph inbox endpoint with a bearer token you provide via:

export OUTLOOK_GRAPH_TOKEN=your_graph_token_here

Then start the continuous worker with:

python main.py watch-inbox \
  --resume-file resume.tex \
  --output-dir artifacts/outlook_jobs

Helpful flags:

• --run-once: poll one time and exit.
• --poll-interval 300: change the interval in seconds.
• --max-messages 20: inspect more recent emails each cycle.
• --extra-context "highlight internships and API work": apply repeated tailoring guidance.

Each job packet folder contains:

• tailored_resume.tex
• tailored_resume.pdf
• application_packet.json
• lead.json

One-off tailoring from files

python main.py tailor \
  --resume-file resume.tex \
  --job-file sample_job.txt \
  --output-pdf artifacts/tailored_resume.pdf \
  --output-tex artifacts/tailored_resume.tex \
  --output-json artifacts/result.json

Frontend flow

Open the site, paste your current resume or LaTeX source, paste the job description, and the app will return:

• a tailored PDF,
• a tailored LaTeX file,
• a concise summary of changes,
• a review reminder before submission.

Notes

• If OPENAI_API_KEY is present, the tailoring flow will try the OpenAI Responses API first and fall back to a deterministic local heuristic if that request fails.
• The prompt explicitly tells the model not to invent experience, credentials, or achievements.
• The Outlook worker currently extracts Handshake links first when present, then falls back to the first URL in the message.
• This repo prepares application materials; it does not promise reliable one-click submission across arbitrary employer sites.