Wojtek Gawroński
Developer Advocate at Unleash (getunleash.io)
Warsaw, Poland
Actions
Wojtek is a software engineer and architect with 15+ years in IT, ex-Amazon, 14x AWS-certified, and co-founder of a cloud-native consultancy that advised CTOs and CIOs across Europe and the US. He has delivered 100+ talks and workshops since 2014. FeatureOps brought him to Unleash: reliable software that solves real user problems, released safely and under control. Now he helps large enterprises adopt an AI-native SDLC while keeping quality and stability high, so they build the right software in the right way, without losing speed or confidence.
Area of Expertise
Topics
The Golden Path to Ship AI-generated Code Safely: Hands-on FeatureOps Workshop
AI is accelerating how fast code gets written. However, DORA research shows delivery stability dropping as AI adoption rises. The missing piece is not better or centrally governed AI. It is better control over what AI produces once it is deployed.
In this hands-on workshop, attendees learn about and experience in practice FeatureOps, an emerging practice and safety net for the AI era of software delivery. Using Unleash and its MCP server, they automatically wrap AI-generated code changes in feature flags, define rollout policies that every change inherits, configure production-signal-driven safeguards that pause or roll back bad releases, and implement lifecycle management to prevent flag sprawl as AI output scales.
Designed for platform engineers building golden paths, this workshop provides a practical, repeatable model for enabling teams to ship AI-assisted code with speed and confidence. Attendees should bring a laptop and their preferred coding assistant to build it live.
Level: 200 (Intermediate)
Category: Platform Engineering: Infrastructure & Delivery
Audience: Software Architects, Software Engineers, DevOps/Infrastructure Engineers
Projects: OpenFeature, Unleash (open source), MCP
Reversibility-as-a-Service: Runtime Control as a Platform Capability
Infrastructure engineers know exactly how to build artifacts and deploy them. However, the decisions that must change between deployments often end up in a ConfigMap and are enforced by a restart. This session proposes a more structured approach to runtime control: put each decision at the layer where it binds, and know what each layer costs to undo. Demos on Kubernetes show Argo CD and Unleash (an open-source feature management platform) working side by side, plus a real distributed system that waits for a safe moment to apply a flag, using OpenTelemetry Baggage and OpenFeature semantic conventions to keep consistency across services. Attendees leave with structured recommendations and best practices they can apply to their platform.
Level: 200 (Intermediate)
Category: Platform Engineering: Infrastructure & Delivery
Audience: Software Architects, Software Engineers, DevOps/Infrastructure Engineers
Projects: Kubernetes, Argo CD, OpenFeature, OpenTelemetry, Unleash (open source)
Flip the Flag, Not the Table: Safe and Controlled AI-powered Releases with FeatureOps
AI assistants and agents write a growing share of production changes. Code reviews and tests cannot keep pace with that volume, and regulators expect AI systems and critical changes to stay under human control. This session treats reversibility as a platform service: every change, from a person or an agent, ships behind a runtime control with a tested way back. Live demos show an agent wrapping its own change in a flag and a rollout that stops when metrics drift, or a kill switch that moves an AI feature to a rules-based fallback, with the audit log recording who pressed it. All of the above are presented as structured recommendations and best practices they can apply to their platform.
Level: 200 (Intermediate)
Category: Agentic AI
Audience: Software Architects, Software Engineers, DevOps/Infrastructure Engineers
Projects: Kubernetes, OpenFeature, OpenTelemetry, Unleash (open source), MCP
Decouple Decision from Execution: Runtime Control for AI Agents
An autonomous agent's loop is a firehose of intentions: it decides it wants to call a tool, API, write somewhere - hundreds of sub-steps a minute. For that, you have to nail the solution to 2 hard questions: is this agent allowed to do that, and should this capability be live now at this stage for those users? Then, if not, can you pull it back in seconds when an agent does something surprising? That's a release-and-reversibility problem, and it's exactly what FeatureOps is for. This talk walks you through FeatureOps deliberate practice, powered by Unleash, that enables reversibility-as-a-service in your Strands Agents deployed on Amazon Bedrock AgentCore with Policy and Gateway.
Level: 300 (Advanced)
Category: Agentic AI
Audience: Software Architects, Software Engineers, DevOps/Infrastructure Engineers
Projects: Kubernetes, Argo CD, OpenFeature, OpenTelemetry, Unleash (open source)
DevOps Won't Save You at Runtime: Enter FeatureOps
Your pipeline is fast. Your observability is mature. Your incident playbooks are tight. And your platform still has a blind spot - in the layer after deployment. This talk hands you a diagnostic for every control decision your platform makes: code, infrastructure, or runtime - and a single question that reveals the gap: when the decision is wrong, how fast can we undo it? Using real 2025/2026 postmortems and the latest DORA report findings, it argues that runtime control belongs on your platform roadmap as a shared capability, not a per-team scramble - especially in the AI-powered era. You'll leave knowing exactly where your runtime-layer gap lies and how to address it.
Level: 200 (Intermediate)
Category: Cloud Native Application Development
Audience: Software Architects, Software Engineers, DevOps/Infrastructure Engineers
Projects: Kubernetes, Argo CD, OpenFeature, OpenTelemetry, Unleash (open source)
OpenFeature Is the Beginning: The Road to FeatureOps
OpenFeature gives teams one vendor-neutral API for feature flags, and for many, it is the first time flags show up in production code. Used well, it acts as an anti-corruption layer: application code depends on the standard, and the flag platform behind it can change. This session shows how far that takes a team and what comes next. Live demos follow one service from file-based flagd to Unleash through the same OpenFeature code, then add the practices and control plane that turn flag evaluation into feature management. Attendees leave with an adoption path and what comes next, and we would like to offer proposals to help OpenFeature grow as an open standard.
Level: 200 (Intermediate)
Category: Cloud Native Application Development
Audience Profile: Software Architects, Software Engineers, DevOps/Infrastructure Engineers
Projects: OpenFeature (incl. flagd, Multi-Provider, OFREP, CLI), OpenTelemetry, Unleash (open source)
WAWTech 2026 Upcoming
Cloud Native Days Poland 2026 Sessionize Event
AWS Community Day Poland 2026 Sessionize Event
PlatformCon 2026
Remote/Online
AWS Community Day Poland 2024 Sessionize Event
AWS User Group Poland (3/2022)
Remote/Online
Please note that Sessionize is not responsible for the accuracy or validity of the data provided by speakers. If you suspect this profile to be fake or spam, please let us know.
Jump to top