Built to Work: Real Engineering in the AI Era
Built to Work: Real Engineering in the AI Era
As AI changes how we build software, reliable systems and teams that learn from experience matter more than ever.
Join us for an evening dedicated to practical lessons on taking a Kafka Connect pipeline into production, building quality into mobile banking from the start, and turning retrospectives into meaningful action.
Drawing on real projects and setbacks, we’ll explore what helps engineering teams improve and where AI can support that work.
This meetup is for software engineers, QA professionals, architects and engineering leaders looking for ideas they can bring back to their own teams.
Agenda:
18:00 – 18:30 – Welcoming participants
18:30 – 18:45 – Welcome thoughts
18:45 – 19:30 – Talk #1 by Bogdan Rotaru
19:30 – 20:00 – Talk #2 by Florian Ostroveanu
20:00 – 20:30 – Talk #3 by Răzvan Dobre
Starting 20:30 – Networking & food & drinks
Talk #1 - From Curiosity to Production: A Kafka Connect Journey at Swissquote
Speaker: Bogdan Rotaru, Senior Software Engineer, Swissquote
Abstract: How a small workgroup turned community-time learning sessions into a production Change Data Capture pipeline
Swissquote's universe of financial instruments is one of the most critical and a frequently changing datasets in the company. Sharing it sustainably with other engineering teams working on self-contained systems and analytics platforms like Snowflake, without handing everyone direct access to the source database, is a real architectural problem.
This is the story of how a few engineers solved it almost entirely in community time: Swissquote's weekly, time-boxed slot for self-directed learning, a small fraction of the hours that go into project delivery. Starting from nothing in those shared hours, we worked our way through Change Data Capture (CDC), Debezium, Apache Kafka, Schema Registry, and Kafka Connect. Starting from first experiments, to a working sandbox and finally to production.
Standing up a Kafka cluster, wiring the pieces into one coherent pipeline, and bending Kafka Connect's configuration model to fit Swissquote's secrets management each took real work, including a custom Kafka Connect config provider we had to write ourselves. A failed first rollout then taught us some hard lessons about WAL growth and connector resilience, and pushed us to build proper monitoring at both the database and connector levels.
What we ended up with is more than the pipeline itself: a pattern other teams can reuse to stream large, fast-changing datasets to many consumers without giving up data ownership or putting the source database at risk and a small piece of evidence for what comes out of giving engineers the time to learn together.
Bio: Bogdan Rotaru is a software engineer interested in the challenges of building reliable, scalable and maintanable software. His experience spans backend engineering, distributed systems and software architecture, as well as leading engineering projects and collaborating across teams and stakeholders. He enjoys understanding how complex systems work and sharing the lessons learned from building them in the real world.
Talk #2 - Building the Foundation, not just the roof - A "shift-left" manifesto and AI context
Speaker: Florian Ostroveanu - Senior QA Automation Engineer, Swissquote
Abstract: In mobile banking, where reliability, security, and speed are critical, traditional end-of-cycle QA is no longer sufficient. This presentation explores how Shift Left QA helps modern mobile teams improve quality by moving testing earlier into the development lifecycle. Using the Swissquote Yuh teams as a practical example, it shows how QA can evolve from a final gatekeeper into a quality enabler embedded in planning, design, development, and automation.
Bio: Senior QA Engineer with 8 years of experience in software quality assurance. Has worked across diverse projects and teams, contributing to quality throughout the software development lifecycle. His areas of interest include Shift left testing, test automation, and fostering collaboration between QA, development, and product teams to build quality in from the start.
Talk #3 - Why retrospectives fail and how to make them useful in the AI Era
Speaker: Răzvan Dobre - Associate Director, Swissquote
Abstract: Retrospectives are easy to schedule, but much harder to make truly useful. The same topics can come up sprint after sprint, action items get forgotten and teams eventually start seeing the retrospective as just another meeting.
AI can now help us summarize discussions, identify patterns and even suggest actions. But can it actually make retrospectives better?
In this session, I'll share some of the patterns I've seen while working with software engineering teams and explore where AI can genuinely help and where it can't. We'll talk about psychological safety, how to get beyond surface-level discussions, ownership, and why following up on what the team decides still matters more than finding the perfect retrospective format or tool.
Because AI can make retrospectives more efficient. It can't make teams more honest with each other.
Bio: Răzvan Dobre is an Associate Director at Swissquote, with a background in software engineering and technology leadership.
He is particularly interested in how engineering organizations evolve, from technical decision-making to team dynamics, ownership and continuous improvement.
This is an in-person event.
Looking forward to sharing insights and ideas with you!