Build a portfolio and host your CV

By the end of this guide you will have a portfolio timeline at /portfolio/, story pages that cross-link your work by tag, and a CV at /cv/ that prints to a clean one-pager and ships a machine-readable cv.json. All from markdown files, all checked by the verifier, all on your own domain.

It builds on your first site; any Cherry site works as the starting point.

How the pieces fit

You write entries into five collections. Cherry renders three surfaces from them, and the cv: block in each entry decides what the third one shows:

1 · The profile

The portfolio's header comes from one file, portfolio.yaml, at the site root:

name: Juno Vale
headline: Grows orchards and software.
location: Belfast
links:
  - label: GitHub
    url: https://github.com/junovale
  - label: Fediverse
    url: https://fedi.example/@juno
updated: 2026-08-01

2 · Your first position

Every collection has a schema, and the schema is the contract. Ask before you write:

$ cherry schema portfolio/positions
portfolio/positions frontmatter:
  title: string (required) — Role title, e.g. "Staff Engineer".
  org: string (required) — Organisation name.
  start: date (required) — Start date.
  end: date — End date; omit while the position is current.
  location: string — City / remote — free text.
  tags: list of string [default: []] — Tags from the shared site taxonomy (cross-linked with blog posts).
  highlights: list of string [default: []] — Short bullet points for the timeline entry.
  cv: cv block — CV curation: `{include, weight, highlights}`. Absent → timeline-only.
  ...

Then write content/portfolio/positions/orchard-systems.md:

---
title: Staff Engineer
org: Orchard Systems
start: 2020-02-01
location: Remote
tags:
  - elixir
highlights:
  - Grew the platform from seed to fruit
  - Led a team of five gardeners
cv:
  include: true
  weight: 10
  highlights:
    - Grew the platform from seed to fruit
---
The long-form story of the orchard years, in markdown. This body
renders on the portfolio page; the frontmatter feeds everything else.

Omit end: while the role is current and the timeline says "present". Note the two highlight lists: the top-level one is for the timeline, the one inside cv: is the tighter cut for employers. More on that in step 5.

3 · Projects and talks, scaffolded

The generators write valid frontmatter so you do not have to remember it:

$ cherry gen.project "Cider Press" --json
{
  "ok": true,
  "command": "gen.project",
  "data": {
    "path": "content/portfolio/projects/cider-press.md",
    "slug": "cider-press"
  }
}

$ cherry gen.talk "Backpressure in practice" --json
{
  "ok": true,
  "command": "gen.talk",
  "data": {
    "path": "content/portfolio/talks/backpressure-in-practice.md",
    "slug": "backpressure-in-practice"
  }
}

Fill in what the scaffold left empty. Projects carry status: (active, paused, archived) and links:; talks carry event:, date:, and optionally video: and slides:. A talk with a recording can embed it right in the body with a video component, facade and all:

::video{youtube="q6Yr9DkTn2k" title="Backpressure in practice"}

The oss/ collection (title, repo, role: author, maintainer, or contributor) and education/ (title, institution, dates) round out the story. The verifier's unfilled-field rule nags about any scaffold string you forgot to replace, which is exactly the nag you want before an employer reads the page.

4 · The timeline and the story pages

Build, and two surfaces exist already:

• /portfolio/ interleaves everything by date under your profile header: positions as ranges, projects with status, talks and education as points in time.
• /story/TAG/ exists for every tag your portfolio shares with your blog. Tag a position elixir and a post elixir, and the story page shows the job and the writing side by side. One taxonomy across the whole site, which is the part hand-rolled portfolios always lose.

5 · Curate the CV

The timeline is your story for peers; the CV is the cut for employers. The cv: block is the whole curation model:

• include: true opts an entry in. No block means timeline-only.
• weight: orders entries within their section, heaviest first.
• highlights: inside cv: overrides the timeline bullets with a tighter set.

/cv/ renders those entries dense and linear: skills backed by years, positions with curated bullets, projects, education. Nothing on it that you did not deliberately include.

6 · Print it, and ship the JSON twin

Open /cv/ and print. The stylesheet strips navigation, forces the light rendition (even from a dark screen, syntax colours included), and lays the page out for A4. That is the one-pager you attach to an application, generated from the same files as everything else.

Beside it, every build emits /cv.json in JSON Resume format, so the machine-readable version of your CV is never out of date with the human one.

7 · Verify like always

$ cherry check --strict
Checked 34 page(s): all clear.

The same rules cover the portfolio: broken links in entry bodies, empty scaffold fields, duplicate titles. Push, and the deploy workflow puts your story on your domain.