LDSK Case Study

Project

Client: LDSK
Industry: AdTech / Digital Signage
Services: QA & Test Automation with Claude, DevOps & CI/CD, Release Management
Engagement: Embedded QA Engineer

LDSK is an adtech startup building media player solutions across Tizen OS, Dise, Grassfish and other signage ecosystems. Their platform includes playlist config and proof of play reporting. Every impression served has to be logged accurately enough to bill against. They needed someone to design the QA process, embed with the playback team and lift quality across the whole development workflow.

Business challenges

Billing grade QA for signage that cannot afford to miss an impression.

Zero structured testing existed across multiple player integrations and signage platforms. Every unverified impression was a potential revenue leak. Devices in signage environments rarely run on ideal networks. Untested CPU and network throttling edge cases were creating silent failures in production.

Samsung VXT CMS screen view with the LDSK Dev workspace and a connected signage display

Goals & objectives

01/Design an end to end QA process for the signage playback team from scratch.
02/Achieve billing grade accuracy on proof of play reporting across all player integrations.
03/Build test coverage that simulates real world conditions including CPU and network throttling.
04/Expand coverage to new player integrations including VXT, ASC and Brightsign as scope grew.

Built with Claude

Claude wrote the test plan and every Playwright test on this project.

One embedded QA engineer covered six signage platforms. Claude did the writing: plans, test code, fixes and failure analysis. The engineer owned the decisions and reviewed every line before it merged.

01
Plan
Claude reads the player specs, the playlist config and the proof of play schema, then drafts a test plan per integration. The engineer edits it and signs it off.
02
Generate
Claude turns the approved plan into Playwright tests, including the CPU and network throttling profiles and the proof of play assertions.
03
Heal
When a player UI or selector changes, Claude rewrites the affected tests and opens the change for review.
04
Triage
Claude reads failed CI runs, player logs and proof of play gaps. It separates real defects from flaky runs before the engineer files a bug.
05
Expand
VXT, ASC and Brightsign each started from the existing suite. Claude adapted the plan and the tests to every new player in days.

Agents plus engineers. Claude writes. A QA engineer reviews every test before it reaches CI.

Playwright automated testing

Playwright suites that run the player the way a real screen does.

Every Playwright suite in LDSK's internal automation repos was written with Claude. The suites run in CI on every change and on BrowserStack for real device and browser combinations.

Throttled conditions

Tests run under CPU and network throttling profiles that match the hardware and connections found in signage locations.

Proof of play checks

Each playback run is compared against the proof of play log. A missing or duplicated impression fails the build.

CI integration

Suites are wired into the release pipeline. Release sign off waits for a green run.

Cross platform

The same suites cover VXT, ASC and Brightsign player integrations through BrowserStack.

// Written with Claude: playback under a constrained device profile
const cdp = await page.context().newCDPSession(page);
await cdp.send('Emulation.setCPUThrottlingRate', { rate: 4 });
await cdp.send('Network.emulateNetworkConditions', {
  offline: false, latency: 400,
  downloadThroughput: 750 * 1024 / 8, uploadThroughput: 250 * 1024 / 8,
});
await expect(page.getByTestId('playback-status')).toHaveText('playing');

Key deliverables

QA Process Design

Built an end to end QA process from scratch for the signage playback team. Claude drafts the test plans and the tests. The engineer runs execution and release sign off.

LiveDOOH player debug dashboard with playback, media and system panels

Network & CPU Throttling

Simulated real world signage conditions where devices run on unreliable networks with constrained hardware. This caught failure modes that never surface on ideal test environments.

Player diagnostics with playlist, media files, errors and proof of plays

Multi Player Integration Coverage

Test coverage across VXT, ASC and Brightsign player integrations. Playback behaves the same on every platform as scope expanded.

Portrait and landscape playback of the same programmatic creative

Proof of Play Validation

Verified billing grade accuracy for every impression logged across the reporting pipeline.

Technologies used

Testing & QA

ClaudePlaywrightBrowserStackManual QACPU ThrottlingNetwork Throttling

Platforms & Integrations

Tizen OSDiseGrassfishVXT PlayerASCBrightsign

CI/CD & Automation

CI IntegrationInternal Automation ReposRelease Management

Process & Collaboration

AgileSlackDocumentationMediation Strategy