AWS Community Day Baltic 2025

In September 2025, the AWS community gathered in Gdynia for AWS Community Day Baltic 2025 – a full-day conference packed with technical talks, live demos, and stories from the world of production systems. The conference focused on practical engineering lessons, showcasing patterns and strategies that teams can use today to build scalable, resilient, and cost-effective systems in the cloud. Participants had the opportunity to see how large organisations integrate new technologies, such as generative AI, into production workflows and how they make architectural decisions at scale.

Model Context Protocol (MCP) & AI in production

The day began with a session on Model Context Protocol (MCP) – a framework that treats AI context as an engineering artifact. MCP transforms raw documents and data into structured, versioned units that are easy to track and orchestrate. This allows agent systems to operate in a deterministic and explainable manner, giving engineers full control over what information is incorporated into AI workflows.

On AWS, you can use: 

• Amazon Bedrock (LLM runtime), 
• Amazon Kendra or OpenSearch (semantic search), 
• Step Functions and Lambda (orchestration), 
• S3 or DynamoDB (context storage), and EventBridge for automatic updates.

The session showed that treating context as a first-class engineering element not only allows for more consistent AI results, but also makes debugging and auditing far easier.

From serverless to functionless architectures

Another topic was simplifying architecture by transitioning from serverless to functionless architectures. The speakers showed that it is possible to completely remove intermediate Lambda functions and connect services directly. 

For example:

• EventBridge can call Step Functions that write data directly to DynamoDB.
• AppSync can communicate with DynamoDB and EventBridge without additional ‘glue’ code.

The benefits are significant: lower operating costs, no cold start delays, and fewer bugs from unnecessary middleware. However, this requires careful planning and an understanding of the dependencies between services. Functional simplicity allows teams to focus on the correctness of business logic and system maintenance rather than boilerplate code.

Event-driven architectures at scale

Speakers also shared examples of event-driven architectures handling millions of images daily. The sessions showed how storage-first principles, fan-out messaging, change logs, and cross-account logging enable the creation of fault-tolerant and horizontally scalable systems.

AWS services played a key role: S3, DynamoDB, SNS, SQS, EventBridge, and Kinesis formed the backbone, while careful selection of partition keys, idempotency, and full system observability ensured reliability. The takeaway: effective scaling requires both architectural discipline and thoughtful use of core AWS services.

Everything as Code

One growing trend was the Everything as Code approach, extending Infrastructure as Code (IaC) practices beyond infrastructure to policies, observability, and security. Representing all operational elements as code enables versioning, automated testing, and full auditability.

CloudFormation and CDK can work with Policy as Code frameworks and observability definitions to create a unified, repeatable, and consistent operating ecosystem. This approach reduces the risk of human error, streamlines the onboarding of new team members, and ensures process consistency across the organisation.

AI agent development

The conversation naturally moved on to the development of AI agents, from local prototypes to multi-agent systems. In local prototypes, the Strands framework allowed roles, memory and tools to be defined, but scaling required shared memory, event-driven coordination and a clear division of roles.

On AWS, production multi-agent systems utilise Bedrock AgentCore, ECS or EKS, CloudWatch and X-Ray, and IAM boundaries for security. By viewing multi-agent AI as distributed systems, teams can achieve high reliability and scalability, creating more advanced and autonomous workflows.

Cloud cost optimization

Cloud costs were the main topic throughout the day. Speakers emphasised that cost reduction is the result of engineering maturity, not ad hoc cuts.

Most fundamental practices include:

• Cost visibility through tagging,
• Control with budgets and anomaly detection,
• Efficiency via right-sizing and autoscaling.

More advanced savings came from workload refactoring or compute lifecycle planning. Crucially, collaboration between finance and engineering ensures cost awareness is built into design rather than forced later.

AWS as an RPG & Well-architected framework

One of the sessions presented the AWS approach as an RPG: each service has its strengths and weaknesses, just like team members in a game. Maintaining a balance between security, reliability, performance, operational excellence, sustainability, and cost requires regular Well-Architected Reviews, which act as checkpoints to identify risks and areas for improvement.

Engineering mindset in the age of AI

Generative AI speeds up coding, but reinforces the importance of engineering practices. Developers must write Architecture Decision Records, use tests-as-specs, and conduct design reviews. AI does not replace discipline, but increases its importance – teams that integrate AI responsibly build more maintainable and reliable systems.

Conclusion

AWS Community Day Baltic 2025 showed that modern cloud engineering is primarily about mindset and discipline, not just technology choices. The lessons covered AI-driven workflows, functionless integrations, event-driven scaling, and cost-conscious design. The tips from the conference provide a blueprint for teams that want to operate effectively on a large scale in the cloud – today and in the future.

Take your cloud engineering to the next level

Inspired by the insights from AWS Community Day Baltic 2025? Whether you’re looking to build scalable, resilient cloud architectures or optimize your AWS costs, SolDevelo can help you implement these best practices in your organization.

• Explore our Cloud Development Solutions to design, build, and maintain production-ready cloud systems tailored to your business needs.
• Discover our AWS Cost Optimization Services to gain full visibility, control, and efficiency in your cloud spending while improving overall system performance.

Let’s turn insights into action, making your cloud architecture smarter, faster, and more cost-effective.