Beyond the Green Checkmark: Clean Code, TDD, and AI Testing with Gil Zilberfeld

What happens when your CI/CD pipeline can’t give you a simple yes or no? Tarek Ziadé, who leads CI/CD on Hugging Face’s Transformers library (and previously built CI at Mozilla), joins David Burns to talk about testing software where the maths changes every time you run it.In this episode:Why green doesn’t mean working: non-determinism, GPU variance and out-of-memory leaks across 500+ model architecturesLiving with 0–500 daily failures out of 13,000 GPU tests, and triaging them with Grafana trends instead of by handSerge, the overnight agent that reproduces failures, builds patches and opens PRs (170 runs, 19 PRs in two months)Catching reward hacking when the agent “fixes” a test by changing the assertionCutting agent costs from $1.50 to under 50 cents a review by moving logic out of prompts and into deterministic PythonA Ruff-style model linter with 58 rules, tiny overfitted models for CPU-only end-to-end checks, and the Model Integration ContractSandboxing agents: separating instructions from data, Bubblewrap, network proxies and a human on the merge buttonPlus the quickfire round: time-to-signal as the most underrated CI metric, why Tarek’s dream job is running coach, and whether AI will ever replace testing.Follow Tarek’s writing on the Hugging Face blog.Subscribe to BrowserStack Talks on Spotify, Apple Podcasts or YouTube so you don’t miss future episodes.

Agile Testing, Whole-Team Quality and AI with Lisa CrispinHow has our understanding of quality changed from the early days of extreme programming, through DevOps, and into the age of AI?In this episode of BrowserStack Talks, David sits down with Lisa Crispin, co-author of the Agile Testing books with Janet Gregory, to talk about whole-team quality and what testers need to thrive next.They cover why DevOps is really just good software development, why small batches still beat big releases, what shift left should actually mean, and how to prioritize risks instead of testing every permutation. Lisa also shares how AI tools can free teams up for more human conversations, how to build a quality culture in remote and hybrid teams, and why communication is the most underrated skill in testing.What you'll learn:- Why continuous integration and small batches are still the biggest hurdles for teams- How to focus on the handful of risks that matter most, and decide whether to test, prevent or monitor them- Where AI adds real value for testing teams, and the trap of forgetting to talk to humans- How feature flags and release strategies make room for exploratory testing- The mindset and skills testers will need over the next five to ten yearsChapters0:00 Welcome to BrowserStack Talks with Lisa Crispin2:30 Whole-team quality: from extreme programming to DevOps5:45 The biggest hurdles: continuous integration and small batches8:11 "I don't have time to test"10:26 What shift left should really mean13:48 Streamlining pipelines and risk-based testing19:09 AI, continuous delivery and the future of testing23:00 Where AI provides genuine value for testing teams27:21 Balancing continuous deployment and exploratory testing30:10 Observability and testing31:17 Quality culture in remote and hybrid teams36:10 Core competencies for testers in the next 5-10 years40:24 Quickfire round49:08 Where to find LisaMentioned in this episode:- Agile Testing, More Agile Testing and Agile Testing Condensed by Lisa Crispin and Janet Gregory- Continuous Delivery by Jez Humble and Dave Farley- Fearless Change and More Fearless Change by Linda Rising and Mary Lynn Manns- DORA (DevOps Research and Assessment) metricsFind Lisa:Website: https://lisacrispin.comLinkedIn: search for Lisa CrispinFollow or subscribe to BrowserStack Talks on your favorite platform so you never miss an episode.

A green build tells you a lot less than you think. Gil Zilberfeld of TestinGil,  trainer and consultant on unit testing, TDD, clean code and API testing,  joins David to unpack what actually makes quality build into software from day one. We cover getting TDD into a ten-year-old codebase that actively resists it, why refactoring is the step everyone quietly drops, what an "80% coverage" target really buys you, and why APIs deserve far more respect than "send a request, check the status code". Then AI: why Gil can't get his coding agent to honour the red step without external scripts and git checks, what self-healing tests really are underneath the magic, and the new ground testers now have to cover — prompt injection, history poisoning, performance, and the token bill. Chapters 00:00 Welcome to BrowserStack Talks 00:31 Meet Gil Zilberfeld, TestinGil 01:06 Video games, flaky tests and testing hype 02:34 Teaching TDD in a legacy codebase that resists it 05:34 Why teams skip the refactor step 11:21 Why TDD matters more in the age of AI agents 16:12 Coverage targets, mutation testing and what really matters 21:03 Good beats perfect: the cost of maintenance 22:30 Giving APIs the respect they deserve 26:59 Complex systems, flaky tests and the elevator analogy 29:27 Breaking flakiness down into something manageable 33:16 Observability and tests as executable documentation 34:57 What do your test results actually mean? 37:09 Self-healing tests: magic, or method? 40:45 Hallucinated edge cases and AI bias in test suites 41:38 Security, performance and tokens: the tester's new remit 44:32 "We don't need QA" and the AI reckoning 48:17 Quick fire round Where to find Guest: Gil Zilberfeld — https://testingil.com Host: David Burns, BrowserStack New episode of BrowserStack Talks every two weeks, where we turn testing pain points into playbooks.

Writing reliable locators is often treated as a rite of passage for automation engineers, and it's just as often the reason pipelines break. In this episode of BrowserStack Talks, host David Burns sits down with Sanjay Kumar, creator of SelectorsHub and TestCase Studio and founder of the SelectorsHub community, to turn testing pain points into playbooks.Sanjay's tools are used by testers around the world every day to save time and cut frustration. We dig into what actually makes element locators fragile, how modern web complexity, Shadow DOMs, nested iframes, and dynamic elements, breaks scripts, and why a clean locator strategy upfront is the difference between scaling your tests and scaling your failures.In this conversation:- The "aha" moment behind SelectorsHub and why beginners had no smart editor for XPath and selectors- Why SelectorsHub generates from the nearest stable element instead of from the top of the DOM- Handling Shadow DOM and iframes that used to cost testers hours of wasted debugging- How TestCase Studio turns recorded actions into plain-English steps — bridging manual testers, analysts, and automation engineers- Self-healing locators and AI hype vs. the discipline of getting the selector right the first time, and why determinism matters- One-click tools for exploratory checks, accessibility, screenshots with URLs, and broken-link scanning- Why development was never the real bottleneck — adoption, maintenance, and human judgment are- A quick-fire round on frameworks, entrepreneurship, and why the world still needs human ingenuity in an AI-driven eraWhether you're a seasoned automation architect or a manual tester moving into low-code automation, this episode is packed with practical takeaways.Explore Sanjay's suite of tools at https://selectorshub.com — and follow Sanjay on LinkedIn and via Testing Daily for the latest from the testing community.Run your scripts across real devices and browsers with BrowserStack to guarantee seamless user experiences.Subscribe to BrowserStack Talks for more conversations that turn testing pain points into playbooks. Until next time — happy testing!Chapters:00:00 Introduction00:58 The Art & Evolution of Element Locators17:00 TestCase Studio & Accelerating the QA Workflow26:58 Self-Healing Locators & the AI Debate38:32 Quick-Fire Round46:13 Closing & Where to Find Sanjay

What if testers developed like engineers — and developers tested like they were in the hot seat? In this episode of BrowserStack Talks, the podcast that turns testing pain points into playbooks, host David Burns sits down with Filip Hric to unpack the philosophy reshaping how we think about quality.Filip has worn just about every hat in the industry — tester, QA lead, test automation engineer, front-end developer, and now Developer Relations Engineer at Qodo, where he works on AI code quality and governance. Together they get into why stepping into the developer's world makes you a sharper tester, how AI has become a genuine learning tool rather than a crutch, why test "flakiness" is really a knowledge gap in disguise, and what quality looks like in an AI-driven, high-velocity world.Along the way: where to start bridging the dev/test divide (hint — API testing and building your own app), the trouble with chasing code coverage, why component testing quietly won, and a rapid-fire quick-fire round to close things out.Whether you're a tester curious about code, a developer who's allergic to writing tests, or a lead trying to make sense of AI in your pipeline, there's something here for you.Chapters 0:00 Intro 0:27 Meet Filip Hric 1:27 Why testers should think like developers 7:35 Where to start: API testing & building your own app 10:48 AI as a learning tool & just-in-time knowledge 14:14 Context windows, agents & the "tester 2.0" 19:51 Fighting flakiness: reframing it as a knowledge gap 24:03 Frameworks, wait libraries & targeting the right element 26:52 Component testing vs end-to-end tests 32:12 The real signal of unstable tests: force-merged PRs 35:30 Why code coverage is a vanity metric 39:19 AI-native vs deterministic testing 40:03 Testing AI-built vs AI-enabled apps 44:29 Security, prompt injection & chatbots 47:39 Quick-fire round 49:08 Where to find Filip & wrap-upConnect with Filip Hric on LinkedIn, Twitter/X, and Bluesky.Subscribe to BrowserStack Talks for more conversations that turn testing pain points into playbooks.

Welcome to Browser Stack Talks, the podcast where we turn testing pain points into playbooks. In this episode, host David Burns sits down with Marie Cruz, Senior Developer Advocate at Grafana Labs and co-author of Contract Testing in Action, to deep dive into the world of performance, reliability, and understanding how systems behave under pressure.Moving away from the traditional model of treating performance testing as a final checkpoint right before production, Marie breaks down how modern software teams can integrate performance thinking into every stage of the development life cycle. Whether you are a developer, QA professional, or SRE, this episode provides a practical playbook for catching critical regressions early without slowing down your deployment pipelines.

Are you tired of chasing empty QA metrics, managing flaky UI tests, and drowning in automation debt? It’s time to move past the tool-driven hype cycle and focus on what actually matters to your engineering pipeline.In this episode of BrowserStack Talks, host David Burns sits down with world-renowned independent test automation consultant and trainer Bas Dijkstra. Together, they strip away the corporate buzzwords to uncover the true essence of modern testing strategy: getting valuable feedback, fast.Bas shares his pragmatic philosophy on building technical confidence, breaking down silos between developers and testers, and why the software world needs to ditch the rigid "tester" archetype in favor of true quality advocacy.What You’ll Learn in This Episode:* The Value Shift: How to transition your team's mindset from tracking superficial "test counts" to measuring actionable feedback value.* Beyond the UI Layer: Why teams still lean on the safety blanket of UI testing, and how to effectively scale with API, integration, and contract testing.* Modern Tooling & Infrastructure: Insights into the rise of Playwright, the creation of Bas’s open-source library RestAssured.Net, and how to scale pipelines seamlessly using cloud infrastructure.* The Reality of AI in 2026: How to responsibly integrate AI into your QA workflows without accumulating piles of unmaintainable, AI-generated technical debt.* The Irreplaceable Human Element: The exact critical thinking and risk analysis skills that will protect a tester's career against complete AI displacement.About the Guest:Bas Dijkstra is an independent test automation consultant, trainer, and developer with nearly two decades of experience helping engineering teams across the globe improve their testing practices. He is the creator of the RestAssured.Net library and the voice behind the widely read blog On Test Automation.Connect with Us:* Follow Bas Dijkstra: Read his latest insights at ontestautomation.com or connect with him on LinkedIn.* Subscribe to BrowserStack Talks: Never miss an episode! Hit Subscribe or Follow on Apple Podcasts, Spotify, or wherever you get your podcasts.* Learn More: Head over to browserstack.com to discover how to scale your testing infrastructure seamlessly.

Can a design mindset completely change how you approach software quality? In this episode of BrowserStack Talks, host David Burns sits down with Brittany Stewart—Senior QA Specialist, AI Lead at QualityWorks, and BrowserStack "Icon of Quality"—to explore the fascinating intersection of creativity, human intuition, and artificial intelligence in modern software testing.Brittany shares her unique journey from graphic and interior design into tech, breaking down how a "design eye" elevates user experience testing. We also cut through the AI hype to discuss the real-world execution of Managed AI Test Squads, how AI drops 15-minute tasks down to under 5 minutes, and why human testers are more vital than ever in the age of automation.Listen now to learn:🧠 How to apply a "design mindset" and mind-mapping to conquer complex test plans.🤖 The practical realities of using AI as a collaborative testing partner, not a replacement.📈 Real-world efficiency gains from AI-driven test case generation.💡 Navigating the "messy parts" of QA, onboarding mid-stream, and building testing communities.Want to be a part of an exclusive community for QA professionals?Join our global QA community on Discord

In this episode of BrowserStack Talks, David and Puja Jaganani, experienced open-source engineers and developer advocates, delve into the essential strategies for starting your test automation journey. Learn why testing should be viewed as an ongoing mindset rather than just a development phase and how to build a solid foundation using the testing pyramid. From picking the right framework to the importance of atomic tests and testing on real browsers, this discussion provides actionable insights to help you gain confidence in your code and accelerate your deliveryWant to be a part of an exclusive community for QA professionals?Join our global QA community on Discord

Join us as Alan Page, a veteran in the tech space with experience spanning Microsoft, Unity, and NBC Universal, shares his insights on modern testing principles, the importance of culture, and leveraging AI in quality assurance. Discover how shifting from traditional testing to a delivery-focused mindset can impact product success and customer satisfaction.Want to be a part of an exclusive community for QA professionals?Join our global QA community on Discord