u11d

Have you ever heard of 211? Across North America, 211 is a free service that connects people with essential community resources - from housing and food assistance to mental health, healthcare, disaster recovery, and many other support programs. Every day, thousands of people rely on it to find the right help. Behind those conversations is one critical asset: reliable data. At u11d, we're the technology partner behind Connect 211 - the modern search engine for community resources. We design, build, and operate the multi-tenant cloud platform that helps organizations manage social service data with high availability and AI-powered workflows. Our collaboration spans the entire platform lifecycle - from architecture and product development to cloud operations, continuous improvements, and ensuring the platform remains resilient and available whenever it's needed. Read more about our work with Connect211: https://lnkd.in/dMPV29uS *** This mission reflects the kind of work we care about most: building systems that support people and communities facing real-world challenges. That idea naturally led us to Deceris - an AI-powered decision support platform that helps organizations predict floods and assess environmental risks, so they can act faster when it really matters for people’s safety and lives. If this is a space you’re interested in, you can follow the Deceris LinkedIn page for updates as we continue building and sharing progress and insights from the project. *** u11d - uninterrupted technology for uninterrupted services #AI #Data #Cloud #HumanServices #GovTech #SocialImpact #DecisionSupport #TechnologyPartner

u11d took part in GITEX AI EUROPE 2026 in Berlin as one of the Polish technology companies presenting their solutions at the Polish National Pavilion. During the event, we showcased Deceris - our operational intelligence platform supporting crisis management and critical infrastructure resilience. Deceris combines AI, geospatial analytics and simulation models, including our proprietary 2D water inundation model, to help assess risks and support operational decision-making. GITEX AI Europe was a valuable opportunity to meet representatives of public administration, infrastructure operators, technology companies and potential partners from across Europe and beyond. We would like to thank the Ministerstwo Cyfryzacji, Polish Investment and Trade Agency (PAIH) and the Urząd Marszałkowski Województwa Zachodniopomorskiego for organizing the Polish National Pavilion and creating excellent conditions for international business networking. The conversations started in Berlin are an important step in further developing Deceris and building partnerships around AI-driven resilience solutions. #GITEXEurope #ArtificialIntelligence #OperationalIntelligence #CriticalInfrastructure #ClimateResilience #Geospatial #GovTech #FloodRisk #DigitalTransformation #Deceris

Animating geospatial time series often means switching between multiple vector tile sources, which can introduce unnecessary network requests, repeated tile decoding, and visible frame flicker. One practical alternative is to package all timesteps into a single PMTiles archive as separate named MVT layers. On the client, playback becomes a lightweight style update instead of reloading tiles, reducing CPU overhead and improving animation smoothness. A useful implementation detail is that advancing a frame only requires updating the active layer reference and triggering layer.changed() in OpenLayers, allowing already-decoded features to be repainted without another fetch. The article walks through the complete approach, including tile generation with Tippecanoe and the OpenLayers implementation: https://lnkd.in/e94hYyuc #PMTiles #VectorTiles #OpenLayers #Geospatial #WebPerformance #GIS

Managing authentication in Argo CD often creates a second challenge: how to provide OAuth credentials without exposing secrets in Git repositories. One approach is to keep GitLab OAuth credentials encrypted in Git using SOPS and age, then decrypt them during reconciliation with KSOPS. This preserves a GitOps workflow while avoiding plaintext secrets in version control. The implementation also highlights an important operational consideration: enabling Kustomize exec plugins for KSOPS functionality should be limited to trusted repositories because of the associated security implications. A detailed walkthrough is available in the article: https://lnkd.in/dC55BxdN #ArgoCD #GitLab #GitOps #Kubernetes #DevOps #SOPS

Organizations frequently over-invest in hosting infrastructure for low-traffic content sites. Landing pages, blogs, or small CMS platforms often do not require always-on compute like VPS or container-based deployments. Serverless architectures address this by shifting compute to a per-request model. In setups such as Payload CMS on Amazon Web Services (AWS) Amplify with MongoDB Atlas as the persistence layer, the application only consumes resources when requests are processed, while the database can also run on a managed free tier. For workloads with low or spiky traffic, this reduces the need to maintain idle infrastructure and lowers baseline operational cost without changing the application architecture significantly. It is a practical option to evaluate for CMS-driven systems where availability requirements are moderate. Read more here: https://lnkd.in/d9VfnwtR #Serverless #PayloadCMS #AWSAmplify #MongoDB #CloudArchitecture #DevOps

Reciprocal Rank Fusion (RRF) is a practical way to combine lexical (BM25) and vector search results without relying on score normalization. Instead of comparing scores from different retrieval methods, it merges results based on their rank positions, which can produce more stable hybrid search behavior across diverse datasets. An interesting implementation detail is that RRF has historically been subject to licensing restrictions in some Elasticsearch deployments, leading teams to either implement rank fusion in the application layer or evaluate OpenSearch, where RRF is available as part of the platform's hybrid search capabilities. For teams building RAG systems or search-heavy applications, retrieval architecture decisions may depend as much on feature availability and operational constraints as on relevance quality itself. Further reading is worthwhile if you're evaluating hybrid search at scale. https://lnkd.in/dPyGKuKZ #Elasticsearch #OpenSearch #HybridSearch #VectorSearch #RAG #DevOps

A recurring issue in serverless architectures is the long first execution caused by cold starts, which can introduce noticeable latency spikes under low or bursty traffic. In Next.js deployments on serverless platforms such as Amazon Web Services (AWS) Amplify, this often shows up as slower initial page loads or API responses during periods of inactivity. One mitigation pattern is to use scheduled triggers via Amazon EventBridge to periodically invoke Lambda functions and keep execution environments warm, separating warmup concerns from application code. This improves latency consistency, but introduces additional infrastructure components that need to be operated and monitored. Worth considering when evaluating performance characteristics of serverless workloads. Read more: https://lnkd.in/dV6-unGr

Last week we participated in AI & Big Data Expo North America in San Jose as part of the business mission organized by the Marshal’s Office of the West Pomeranian Voivodeship. The event brought together a very broad cross-section of the AI ecosystem, from enterprise AI platforms and infrastructure providers to practitioners working on real-world deployment challenges. A few observations stood out particularly strongly: 👉 the discussion is shifting from “AI demos” toward production-grade systems and operational reliability, 👉 orchestration, governance and data quality are becoming central topics in enterprise AI adoption, 👉 companies are increasingly focused on domain-specific AI systems rather than generic models, 👉 infrastructure, observability and operational resilience are becoming just as important as the models themselves. For us, it was also a valuable opportunity to discuss topics closely related to our work in data engineering, geospatial systems and operational intelligence platforms. We also had the opportunity to present deceris.io during a networking session in Silicon Valley and exchange perspectives with companies and organizations operating on the US market. Participation in the event was co-financed under the “Ster na Eksport” project implemented within the European Funds for West Pomerania 2021–2027 programme. #AI #DataEngineering #AIInfrastructure #BigData #OperationalIntelligence #Geospatial #SanJose #TechEx

We are pleased to announce that u11d has received support under the “Ster na Eksport” project for participation in GITEX Europe Berlin 2026 and international export development activities. During the event, we will represent Poland and the West Pomerania region, presenting our work in operational intelligence systems, AI, geospatial analytics, and critical infrastructure technologies. Participation in the event is co-financed under the non-competitive project “Ster na Eksport”, implemented within the European Funds for West Pomerania 2021–2027 program, Measure 1.9 “Preparation and implementation of new business models for enterprises”. #GITEX #AI #OperationalIntelligence #Geospatial #CriticalInfrastructure #BigData #Cloud #Poland

Keycloak's built-in email domain restriction uses a static allow-list tied to realm configuration — updating it requires a config change and redeploy, which becomes a real bottleneck in multi-tenant B2B environments. A cleaner approach is a custom Authenticator SPI that delegates the decision to an external service at login time. The contract is minimal: Keycloak sends { domain, realmId } via HTTP POST, and the backend responds with 200 to allow or any non-200 to deny — keeping all tenant policy logic in your backend, not in Keycloak config. The validator is fail-closed by design: network errors, timeouts, and misconfigurations all result in denial, not bypass. The same authenticator class handles both username/password browser flows and post-broker IDP flows (e.g. Google SSO), configured per flow execution through the Keycloak Admin UI. The full article covers the Java SPI implementation, Docker multi-stage build, flow configuration steps, and operational tradeoffs of this pattern: https://lnkd.in/d_dp4bvB #Keycloak #IAM #DevOps #Authentication #MultiTenant #BackendEngineering