Dossier

My work history, automatically turned into professional documents.

Dossier renders my resume from a YAML source of truth and publishes it via a Cloudflare Worker.

How it works

flowchart LR
    classDef source fill:#fde68a,stroke:#b45309,stroke-width:2px,color:#1f2937
    classDef stage  fill:#bae6fd,stroke:#0369a1,stroke-width:2px,color:#0c4a6e
    classDef artifact fill:#bbf7d0,stroke:#15803d,stroke-width:2px,color:#14532d
    classDef deploy fill:#fbcfe8,stroke:#be185d,stroke-width:2px,color:#831843
    classDef edge   fill:#ddd6fe,stroke:#6d28d9,stroke-width:2px,color:#3b0764

    subgraph S[sources/resumes/]
        TECH[SDE2_CV_v2.yaml<br/>tech]:::source
        IMPACT[SDE2_CV.yaml<br/>impact]:::source
    end

    R([render]):::stage
    RA([render-anon]):::stage
    PP([publish-pdf]):::stage
    OG([og-image]):::stage
    DW([deploy-worker]):::stage

    ANON[SDE2_CV.anon.<br/>generated.yaml]:::source
    PDFS[artifacts/resumes/<br/>PDF + PNG + Typst]:::artifact
    PUB[deployments/cf-workers/<br/>public/*.pdf]:::artifact
    OGIMG[public/og-image.png]:::artifact

    CF{{Cloudflare Worker<br/>worker.js + pdf.js}}:::deploy
    WEB((resume.vibhakar.dev<br/>resume.vibhakar.in)):::edge

    TECH --> R
    IMPACT --> R
    IMPACT --> RA --> ANON --> PDFS
    R --> PDFS
    PDFS --> PP --> PUB
    PDFS --> OG --> OGIMG
    PUB --> DW
    OGIMG --> DW
    DW --> CF --> WEB

Render — render runs RenderCV on both YAML variants (tech + impact); render-anon produces a redacted variant for public feedback.
Stage assets — publish-pdf copies the PDFs into the Worker's public/ directory; og-image builds a 1200×630 social preview from the rendered PNG.
Deploy — deploy-worker ships public/ via wrangler to a Cloudflare Worker that serves the PDF directly (with SEO + OG tags) and a vendored pdf.js viewer for browsers.

Documents generated

Resume (1 page)

Concise and ATS-friendly, rendered from sources/resumes/SDE2_CV.yaml.

Anonymized resume

A redacted version (header + socials stripped) for public feedback, written to SDE2_CV.anon.generated.yaml.

Tech stack

Document generation — RenderCV for the resume
Hosting — Cloudflare Worker serving the PDF and pdf.js viewer
Web viewer — Vendored pdf.js for in-browser PDF display

Build stages

The pipeline lives in dossier.py. Run everything with uv run dossier.py, or pick a subset with uv run dossier.py <stage> [<stage> ...]. List stages with uv run dossier.py --list.

Stage	What it does
`render`	Renders both resume variants (tech + impact) to PDF + PNG via RenderCV, into `artifacts/resumes/`.
`publish-pdf`	Copies both rendered resume PDFs into `deployments/cf-workers/public/` for the Worker to serve.
`render-anon`	Writes `SDE2_CV.anon.generated.yaml` with header + socials redacted, then renders it.
`og-image`	Builds a 1200×630 `og-image.png` from the RenderCV PNG output.
`deploy-worker`	Deploys the Cloudflare Worker (publishes `public/` assets to `resume.vibhakar.{dev,in}`).

Project structure

dossier/
├── dossier.py                            # Pipeline entrypoint
├── pyproject.toml
├── uv.lock
├── sources/
│   └── resumes/
│       ├── SDE2_CV.yaml                  # Source of truth for the resume
│       ├── SDE2_CV.anon.generated.yaml   # Written by `render-anon`
│       └── fonts/
├── artifacts/
│   └── resumes/                          # RenderCV output (PDF, PNG, Typst)
├── deployments/
│   ├── cf-workers/
│   │   ├── worker.js                     # Cloudflare Worker (SEO, OG tags, viewer routing)
│   │   ├── wrangler.toml
│   │   └── public/
│   │       ├── vibhakar-solanki-sde2-resume.pdf  # Published by `publish-pdf`
│   │       ├── og-image.png              # Built by `og-image`
│   │       └── pdfjs/                    # Vendored pdf.js viewer
│   └── scripts/
│       └── build_og_image.py
└── static/
    └── artifact_pipeline_diagram.svg

Setup

git clone https://github.com/your-username/dossier
cd dossier
uv sync

Run the full build pipeline:

uv run dossier.py

Run a subset of stages (e.g. re-render and publish after a YAML edit):

uv run dossier.py render publish-pdf

List available stages:

uv run dossier.py --list

ThingsToAdd

Features described in earlier drafts but not yet implemented:

Git collector — Walk every repository in /projects, pulling commit messages and diffs to feed downstream generators.
Linear collector — Fetch issues I worked on via the Linear API for task/cycle/project context.
Context synthesis — Combine raw git + Linear data into structured context (what shipped, problems solved, collaborators).
Brag document generator — 5–10 page self-review covering goals, projects, collaboration, mentorship, design/docs, company building, learning, and outside-of-work highlights. Intended to feed the resume/CV generators as narrative context.
CV (multi-page) — Full career record with detailed project descriptions, technologies used, and measurable outcomes. Currently only the 1-page resume is generated.
GitHub Actions CI/CD — Scheduled (e.g. weekly) and push-triggered workflow that runs collectors + generators and commits PDFs to a gh-pages branch.
GitHub Pages hosting — Serve the pdf.js viewer from GitHub Pages pointing at the latest PDFs. (Currently hosted via Cloudflare Worker instead.)

License

MIT

MagneticNeedle/Dossier

README