Tiger Kaovilai

Senior Software Engineer at Red Hat, OADP maintainer and Velero maintainer. Open source contributor focused on cloud-native data protection for Kubernetes.

Project maintained by kaovilai Hosted on GitHub Pages — Theme by mattgraham

Data files and workstream dashboard

Reference for the machine-generated JSON in this repo (moved out of the profile README).

career-map.json — curated career connections

Unlike activity data, this file is manually curated from the profile’s existing career and education facts. No resume scraping, automated dates, or inferred employment duration. Update the README’s expandable text version alongside it; tests compare each fallback row’s names, qualifications, roles, contexts, focus, place details, dates, and tools against the curated source.

Field Contents
places[] Ordered schematic route: id, city, country, summary, schools[] (name, nullable dates), skills[]
education[] NC State degree, NC State minor, CS50: name, qualification, dates, skills[]
industry[] Red Hat engineering, Red Hat marketing internship, Deutsche Bank internship: id, name, role, context, focus, skills[]; employment dates unspecified
community[] Velero maintainer: id, name, role, focus, skills[]
projects Personal website / delivery work: name, focus, skills[]
links Profile resume, LinkedIn, OADP, and Velero URLs
note Provenance and limits displayed in the detailed view

Array order is the curated presentation layout, not a measured timeline. The renderer expects the three places, three education entries, three industry entries, and one community entry above; adding a new branch requires a layout change rather than silently hiding it.

.github/scripts/career_map.py renders desktop/mobile SVGs in both themes and the interactive /career/ Pages companion, including a full text table. All outputs are generated through .github/scripts/generate-profile-svgs.py and committed by the existing daily workflow. README and no-JavaScript animations are CSS-only and finish within four seconds. The Pages view adds optional keyboard/hover detail and theme selection; continuous motion starts only after its pause control is wired and visible. Reduced-motion preferences disable all animation.

python3 -I -B .github/scripts/test_career_map.py

open-prs.json Schema

open-prs.json is automatically generated on the update-pr-badges workflow schedule and exported for consumption by kaovilai.pw.

Top-level fields

Field Type Description
updatedAt string (ISO 8601) Timestamp of last generation
prs array All open PRs by kaovilai (sorted by org then status priority)
reviewQueue object Professional review-queue panel data

prs[] item fields

Field Type Description
number number PR number
repo string owner/name
org string Repository owner/org login
title string PR title
url string PR URL
targetBranch string Base branch
status string Badge status: ready, waiting-merge, ci-pending, draft, stale, hold, failing-ci, needs-attention
milestone string|null Linked issue milestone (velero repos only)
author string GitHub login of PR author
isCopilotAuthored boolean True when authored by a Copilot coding-agent bot
assignees string[] Assignee logins
isDraft boolean Draft state
labels string[] Label names
updatedAt string (ISO 8601) Last activity timestamp
workstream string Default workstream classification (Velero — includes the official Velero plugin repos, OADP, KubeVirt Data Mover, Kubernetes — kubernetes/*/kubernetes-sigs/*, e.g. Prow, kubectl, CNCF Landscape, or Uncategorized) — see workstream-classification.json for overrides
ciChecks array Normalized [{name, conclusion}] CI checks from statusCheckRollup, with merge-queue noise (Tide, Auto Request Review) filtered out

reviewQueue fields

Includes only public professional contributions (repos owned by GitHub organizations, not personal accounts). Excludes drafts, rebase-blocked PRs, and merge-conflicted PRs.

Field Type Description
updatedAt string (ISO 8601) Timestamp of last generation
needsReview array Open PRs still waiting for human review/approval, sorted by longest waiting first
approvedWaitingToLand array Sufficiently approved PRs still open (CI, merge queue, hold, etc.), sorted by longest waiting first

reviewQueue.needsReview[] / reviewQueue.approvedWaitingToLand[] item fields

Field Type Description
number number PR number
repo string owner/name
org string Organization login
title string PR title
url string PR URL
targetBranch string Base branch
author string GitHub login of PR author
isCopilotAuthored boolean True when authored by Copilot coding-agent bot
assignees string[] Assignee logins
labels string[] Label names
mergeStateStatus string GitHub merge state: CLEAN, DIRTY, BEHIND, BLOCKED, UNKNOWN, etc.
isApproved boolean True if sufficiently approved (reviewDecision is APPROVED, OR both lgtm+approved Prow labels, OR an approving review when no reviewDecision is reported)
hasProwApproval boolean True if both lgtm and approved labels are present
hasGithubApproval boolean True if at least one native GitHub approving review exists
reviewDecision string|null GitHub PullRequest.reviewDecision: APPROVED, REVIEW_REQUIRED, or CHANGES_REQUESTED (null when reviews are not required). Honors the required approving review count wherever it is configured (direct branch protection or the openshift/release Prow branch-protector)
approvalCount number Number of latest reviews in the APPROVED state (one per reviewer)
requiredApprovals number|null Required approving review count from branch protection rules, falling back to the openshift/release Prow branch-protector config; null when unknown
reviewRequests string[] Requested reviewer/team logins
group string needsReview or approvedWaitingToLand
reason string Classification reason: awaitingReview, hold, pendingMerge
waitingDays number Days since last activity
updatedAt string (ISO 8601) Last activity timestamp

Validation

bash .github/scripts/validate-open-prs-json.sh open-prs.json

open-issues.json Schema

open-issues.json is generated by the update-issues-list workflow (same schedule as PR badges), mirroring open-prs.json for issues: authored, assigned, commented-on (last 90 days), and — for openshift/openshift-eng specifically — anything I’m otherwise involved in (mentioned, review-requested), since those are my employer’s orgs.

Field Type Description
number, repo, org, title, url, labels, updatedAt, workstream — Same meaning as open-prs.json
status string open or stale (no activity in 60+ days)
author string Issue author login
relation string Why this issue is tracked: author, assignee, commenter, or involved
assignees string[] Assignee logins

Validate with bash .github/scripts/validate-open-issues-json.sh open-issues.json.

workstream-archive.json Schema

Append-only historical catalog of closed/merged PRs and issues, going back to the start of my public GitHub activity (not just a rolling window). Once an item lands here it’s treated as a immutable snapshot — the incremental refresh (update-history-archive workflow, daily) only re-touches an entry if it reopens or gets a new comment after close; quiet closed items cost zero API calls on every future run.

Field Type Description
generatedAt string (ISO 8601) Last refresh timestamp — also the cursor the incremental refresh queries --updated=">=" from
items[] array {number, repo, org, title, url, type: "pr"\|"issue", state: "closed"\|"merged", createdAt, closedAt, updatedAt, workstream}

Seeding a fresh archive (or re-seeding from scratch) is manual, not scheduled:

bash .github/scripts/backfill-history.sh   # one-time, all-time, windowed by year

workstream-classification.json / workstream-layout.json

Manual override layer for the /workstream/ dashboard — never touched by the scheduled data-collection workflows.

/workstream/ — Workstream Tree dashboard

A vertical timeline across Velero (including its plugin repos), OADP, KubeVirt Data Mover, Kubernetes (kubernetes/*/kubernetes-sigs/* — split from CNCF Landscape since it’s Kubernetes-specific work, not general CNCF-wide), and CNCF Landscape work: “now” at the top, scroll down for history (same px-per-day scale throughout; pinch/ctrl+scroll to zoom), colored + iconed by status (never color alone — colorblind-safe palette), with CI status per PR and a freshness banner since that data is only as fresh as the last scheduled run.

To review and annotate locally, with edits committed and pushed so they show up on the live site:

bash workstream/serve.sh            # starts a local server at http://localhost:8420/workstream/
bash workstream/refresh-data.sh     # pulls fresh data on demand, between scheduled runs

Click a card to add a note or override its workstream; drag to reorder within a lane or drop onto a different lane to reclassify. “Save & Publish” writes workstream-classification.json/workstream-layout.json and runs a signed-off git commit + push via the local server — so it only ever touches those two files, never the auto-collected data. Without the local server running, edits fall back to browser localStorage only.

Local server security model

“Save & Publish” commits and pushes with your credentials, so workstream/server.py only accepts requests from the dashboard it serves itself: no CORS, loopback Host/Origin only, a JSON content type, and a random per-run token that only a same-origin read of /api/health returns. It serves only workstream/ and the dashboard’s JSON inputs (never .git/), commits just the two annotation files, and refuses to publish unless you are on main with nothing else unpushed. Tests: python3 -I workstream/test_server.py.