Skip to content
Part of ELO — Embedded Logic Operations · Scope, test architecture & API definitions · dev-team lead ↗

In use at Hyundai California

Optimo

Service-automation testing framework for Hyundai's connected-car ecosystem in California — one spreadsheet test script compiles to JMeter, Appium, and Selenium on the fly, and voice journeys are exercised through simulated Alexa / Google Home conversations.

Role
Scope, test architecture & API definitions · dev-team lead (via ELO)
Period
Multi-year engagement · in production use
Stack
Go · Angular · Python · Docker · and 6 more
Place
Hyundai HATA · California, US (Remote)
  • Hyundai HATA

    One spreadsheet script drives three test engines for Hyundai's connected cars.

Test Cases — IoT run configuration: functional/performance reporting, API/voice/speech automation, Google or Amazon device, staging/production/QA, per distributor.

Test Cases — IoT run configuration: functional/performance reporting, API/voice/speech automation, Google or Amazon device, staging/production/QA, per distributor.

HyundaiHATA · California

A single connected-car journey — "Alexa, ask Blue Link to remote start my car" — crosses a voice assistant, an IoT platform, several service back-ends, and the telematics unit in the vehicle itself, then comes back the same way. Any hop can be the one that breaks, and none of them can be verified from a browser.

01 / 03

Why a connected car is hard to test

Hyundai needed that whole chain regression-tested continuously, across distributors and environments, by a QA organisation where not everyone writes code. Optimo was built to be that instrument: a single framework where a test author configures a run — functional or performance, API / voice / speech, Google or Amazon device, staging / production / QA, per distributor — and the platform generates and drives the underlying scripts.

Test Runs — a completed voice-triggered run traced hop by hop (IoT platform → service back-ends → telematics unit and back) with per-step latency.
02 / 03

What Optimo does

  • Human–IoT conversation simulation. Alexa and Google Home dialogues driven as test steps, so the voice services offered to customers are exercised the way a customer actually reaches them.
  • Excel-driven test authoring. A customised pseudo-syntax lets QA write test scripts in a spreadsheet; Optimo generates JMeter, Appium, and Selenium scripts on the fly from it.
  • Step-timed run traces. Every request/response hop between the IoT platform, the service back-ends, and the vehicle telematics unit is recorded in order with its own latency, so a failure is attributed to a hop rather than to "the system".
  • Machine monitoring and load balancing. A distributed-system management tool tracks the test machines and spreads load across them.
  • Scheduled runs and reporting. Automated scheduled testing with step-by-step report generation and PDF emailing to the stakeholders who never open the tool.
03 / 03

Where I sat

  • Early scope definition. Shaped what the framework had to cover before it was built — which surfaces, which environments, which distributor permutations.
  • Test cases and workflows. Wrote test cases and designed the test workflows the spreadsheet syntax had to express.
  • API definitions. Defined the API contracts between the framework, the IoT platform, and the service back-ends under test.
  • Development team management. Ran the engineering team delivering the platform against a client QA organisation as the counterpart.
  • Generated-script framework and autorun planning. Built the testing of the scripts Optimo generates, and planned how runs execute automatically.
  • Excel to JSON for the business team. Complex API and JSON structures behind an Excel import, so the business team could bring in their own workflows and run them.

Decisions and why

  1. Spreadsheet as the test-authoring surface

    WhyThe people who know the connected-car service behaviour are not all script authors. Making Excel the input and generating JMeter/Appium/Selenium underneath kept authorship with the domain experts instead of bottlenecking on an automation team.

  2. One framework across IoT, web, and mobile

    WhyThe failure modes worth catching live in the seams between surfaces. Three separate tool stacks would have tested three halves of the journey and none of the joins.

  3. Step-level timing captured per hop

    WhyConnected-car regressions usually present as "sometimes slow", not "broken". Recording each hop with its own latency turns that into an attributable number a team can act on.

Technical highlights

  • 01

    Spreadsheet pseudo-syntax compiled to JMeter, Appium, and Selenium scripts at run time.

  • 02

    Voice-assistant conversation simulation (Alexa / Google Home) as first-class test steps.

  • 03

    Per-hop request/response tracing across the IoT platform, service back-ends, and vehicle telematics unit.

  • 04

    Run configuration matrix across reporting type, automation type, device, environment, and distributor.

  • 05

    Machine-monitoring and load-balancer tooling for distributed test execution.

  • 06

    Scheduled runs with step-by-step reports and automated PDF distribution.

Impact

  • Gave Hyundai California a repeatable way to test the online architecture behind their connected-cars ecosystem — still in use per ELO's company profile.
  • Moved test authoring to spreadsheet-literate QA staff, removing the automation-engineer bottleneck.
  • Made voice-assistant journeys testable at all, rather than verified by hand on a device.
Outcomes

By the numbers.

3 surfaces
IoT · web portal · mobile app
1 script → 3 engines
JMeter · Appium · Selenium
Voice + API
Alexa / Google Home simulation
↗ Next case study

Jamahook

Jamahook AG is a Zürich-based music-tech company and a direct ELO client for over 6 years — the world's first AI-powered music matching ecosystem, launched in October 2020, with a desktop agent, VST/AU/AAX plugins, and a web portal. Producers find complementary material through acoustic and psychoacoustic audio annotation and analysis, on an algorithm developed in collaboration with Fraunhofer IDMT. I joined in October 2019 to get that matching engine running, led the build of the web platform and admin, launched in 2020, then stood up the full-stack team (React + Node/NestJS on an NX monorepo) and stayed hands-on through the first market release. The dedicated team I helped build still ships modern integrations; since 2023 my seat has been primarily client and business relationship. In November 2025 Jamahook partnered with Matt Pelling to buy Loopmasters and Loopcloud back from Beatport, and in September 2026 Loopmasters, the company behind Loopcloud, acquired Jamahook. The matching engine has run inside Loopcloud since Loopcloud 6 in April 2021.