I am a quality engineering leader moving QA from manual test-and-log execution to an AI-augmented practice, where the engineer works as a prompt engineer and the AI does the legwork. Ten-plus years in QA automation, test and release ownership, and quality strategy.
My point of view on where quality engineering is going.
I do not think QA's future is more people reading logs by hand. It is quality engineers working alongside AI, designing the prompts and correlation logic that read failures at scale, then validating what the AI proposes. I have applied this hands-on to a live, multi-release SaaS platform and quantified its impact in both engineer-hours and dollars. The human always stays in control: AI proposes with a confidence score and the evidence, the engineer decides.
The work I am most known for, and where I want to lead a team next.
Initiated and built a working AI failure-triage proof of concept (Python log parsers, failure analyzers, pipeline integrations) plus a target-state dashboard that clusters failures by likely root cause with correlated logs and deployment signals, so an engineer starts from a probable cause instead of a blank page.
Quantified the case and pitched it to the CEO and senior leadership and at a UK All-Hands, framing the move from tester to quality and prompt engineer.
Used AI as a daily force multiplier across defect analysis, pipeline RCA, coverage analysis, and stakeholder deliverables, with OKR-style hypotheses to measure impact on triage time, RCA turnaround, and accuracy, always validated human-in-the-loop.
Applied AI to coverage strategy: validated an AI gap-finder across four services, correcting false positives and separating QA-ownable work from dev dependencies.
A few pieces of work that show how I lead on quality.
When the regression suite failed, triage meant opening several tools and reconstructing the root cause by hand, which took a long time per failure and multiplied across every environment and parallel release.
Built a working proof of concept in Python that pulls the failures, correlates them with logs and deployment signals, and clusters them by likely root cause. Designed the prompts and correlation logic; kept the engineer in control as the validator.
Failure triage moves from manual log-reading to a probable cause with evidence up front. Quantified the case and pitched the AI-SDLC shift to the CEO and senior leadership, and demoed it at a UK All-Hands.
The regression automation pipeline was down for about a month and produced no results. On the surface it looked like flaky tests, which risked the team spending weeks chasing failures that were never real.
Ran a full root-cause analysis across hundreds of console logs, reading JVM crash codes, memory access violations, and job-lock timing. Traced it to an infrastructure and memory issue on the build agent, applied the fixes, consolidated environments, and lined up the permanent fix with DevOps.
A pipeline blocked for about a month was producing results again, and the team stopped chasing phantom test failures. The infra-versus-test distinction was the whole game.
The existing tool carried heavy maintenance overhead, a paid license, and a Windows build dependency that contributed to the outage above.
Ran the feasibility spike, drove framework selection (REST Assured with JUnit 5), and authored a phased migration roadmap with per-service effort sizing, a CI integration plan, and clear retirement criteria for the old tool.
A costed, agreed path to a code-based framework that runs on any Linux CI agent, reducing fragility and maintenance cost.
Ten-plus years across SaaS, web, and desktop quality.
What I bring to a QA function.
I am looking to lead a QA function into the AI-SDLC era. If that is where you are headed, get in touch.