SenkronAI – Case Study
Senkron AI is a healthcare technology company building Chat365, a chatbot platform that serves enterprise healthcare customers, delivering privacy-sensitive healthcare conversations at scale. Limited by its existing infrastructure and its ability to grow, Senkron AI engaged Onkatec to migrate Chat365 to AWS and establish a secure, cloud-native foundation for its future.
Customer Challenge
The migration was shaped by three challenges. First, the existing infrastructure could not scale to meet the demands of enterprise healthcare customers, lacking the operational tooling required for a production-grade platform. Second, as a healthcare workload, the platform required data protection and compliance to be embedded in the architecture itself — encryption, centralized access management, and PII safeguards — rather than added afterward.
Third, the multi-agent design demanded capabilities that unmanaged infrastructure could not provide: durable event streaming for real-time agent communication, and managed foundation-model access to replace external GPU and LLM dependencies. Each of these needed to be delivered reproducibly and securely.
Partner Solution
Onkatec delivered a production-grade, cloud-native platform defined entirely in Terraform across separate staging and production AWS accounts. At its core, Amazon EKS runs the containerized Chat365 platform on Kubernetes 1.35 with private endpoints and Graviton worker nodes across three Availability Zones, with Horizontal Pod Autoscaling applied to stateless services.
Delivery is automated through GitHub Actions, with self-hosted runners managed by the Actions Runner Controller on EKS. Authentication uses a GitHub App at the organization level rather than personal tokens, and every image is versioned by commit SHA — never latest — for full traceability and rollback. Build and deployment are deliberately separated: images are built and pushed to Amazon ECR, while deployment to the private cluster runs through ephemeral runner pods scoped by Kubernetes RBAC.
For messaging, Strimzi Kafka runs on the EKS cluster in KRaft mode across three Availability Zones, with pod anti-affinity, topology spread, and rack awareness for resilience. Kafka traffic is secured with TLS, SCRAM-SHA-512 authentication, and ACL-based authorization, and a PodDisruptionBudget keeps the cluster available during maintenance.
The data layer uses Amazon RDS for PostgreSQL for transactional and vector data, and Amazon ElastiCache for Valkey for caching, both encrypted. For AI, Amazon Bedrock provides foundation-model access with pod-level IAM permissions via IRSA/EKS Pod Identity, integrated with Langfuse observability and a PII vault for healthcare data protection.
Observability is centralized on Grafana Cloud, with Loki and Alloy for logs and OTLP telemetry alongside kube-state-metrics, cAdvisor, and node-exporter. AWS Secrets Manager, synchronized into the cluster through the External Secrets Operator, centralizes credential management, while Amazon S3 and CloudFront serve the platform’s storage and edge needs.
Results and Benefits
Chat365 now runs in production on EKS, with its API, scheduler, and six worker services independently deployable through per-service manifests, ECR repositories, and Secrets Manager entries. The GitHub Actions pipeline delivers commit-SHA-tagged images to production daily, with manual service selection providing a controlled release path for the multi-service codebase.
The Kafka layer delivers durable, replayable event streaming across three Availability Zones, with event and tenant-configuration topics and a PodDisruptionBudget ensuring the cluster tolerates the loss of a single node. Amazon Bedrock integration gives the platform managed AI access with pod-level permissions and PII protection.
Observability was a particular focus. By filtering high-cardinality metrics and adjusting scrape intervals, Onkatec reduced the active metric series by roughly 55% — cutting projected Grafana metrics spend from around $500 to $70 per month while keeping full visibility across staging and production.
About the Partner
Onkatec specializes in Cloud DevOps and infrastructure optimization, delivering SLA-driven solutions on AWS for organizations with mission-critical workloads. With proven AWS expertise and a team of certified professionals, Onkatec offers comprehensive services across cloud migration, infrastructure as code (IaC), CI/CD pipelines, observability, and cloud-native development.
Our AWS-validated capabilities ensure secure, scalable, and high-performing cloud environments tailored to customer needs. Through expert-led automation, proactive support, and architectural best practices, we help organizations reduce cloud complexity and accelerate digital transformation. Onkatec has consistently delivered measurable business outcomes, such as 30–50% reductions in IT operational costs, improved uptime, and faster time-to-market.
