Software Quality Conference | QLTY PULSE Lithuania

10 steps of project modernization

​

Modernizing a project is a long journey. This presentation is about one of those trips—where it started, which steps mattered most, and the challenges encountered. So come along, and let's explore how quality can guide the way.

Quality Assistance in action at Kayak    

We will go over the quality strategy of travel search website Kayak and how the Quality Assistance model is applied in real life. We will talk about how to be successful as a quality advocate in this setup, share details on topics like test coverage measurement, quality metrics, AI adoption, exploratory testing practices, shift left and shift right practices.

Can you trust your AI notes? Scaling QA for AI Features in Communications

​

As AI-driven features such as real-time transcriptions, captions, summaries, and translation become a standard in communication platforms, traditional QA methods are no longer sufficient. This talk explores the challenges and benefits of scaling the testing of these features using standardized and industry-recognized methods, together with LLM assistance. Ieva will share insights into the overall methodology, the complexities of working with the LLM-as-a-judge method, and how proper planning, preparation, and a mix of metrics can help build trust in the results.

Taming the Zoo: Building a Device Farm That a Company Actually Uses

​

What started as a need to test on real devices grew into a unified testing platform — and the journey is still ongoing. This talk covers the lessons learned so far: from early architecture decisions and getting buy-in, to the challenge of making other teams actually want to use what you've built. Honest takeaways on building internal tooling when adoption is never guaranteed.

Testing AI with AI: Building a Custom Agentic Evaluation Framework

​

Building agentic application can be complicated - but keeping it consistent and accurate takes that complexity to the whole new level. 

A Fortune-teller Told Me: My Rollercoaster Ride Through Work, Life, and AI

​

"Work… well, that's also going to be like a rollercoaster." And so there I was, an opinionated skeptic, unknowingly living out the prophecies of fortune tellers. Who knew that I'd even have a phase of being genuinely hyped about AI, but, like all the highs of rollercoasters, it wouldn't last too long...

Quality at scale: how do we ship 1600+ experiments annually and trust the results

​

Delivering a user-centric product requires rapid experimentation, but how do you maintain stability when 100+ different versions of your product are live simultaneously? This session explores the engineering rigour required to ship 1,600+ annual experiments without compromising velocity and trust in data.

Let metrics do the talking

​

I didn't climb the ladder. I measured my way up.

Clarity when producing Play

What happens when design, engineering, data, business and Fun all pull in different directions - but the game still has to ship? After 8 years of making games Povilas has one answer: clarity is the job, and even when it feels like a losing battle, you have to protect it.