Success Story
Kubernetes modernization and FinOps for Emi Labs: −98% redundant load balancers and costs under control
Emi Labs is an HR Tech platform focused on recruiting and hiring, with several core applications running high traffic volumes and handling critical candidate data. Craftech has supported Emi Labs since late 2023 as its platform team on AWS, covering modernization, FinOps and database modernization.
−98%
redundant network resources (80+ load balancers → 1 per environment)
>US$100/mo
savings per optimized database server
Challenge
Emi Labs ran on a rigid, costly infrastructure that mixed test and production environments, constantly putting real business data at risk. Limitations in the previous architecture meant paying for dozens of redundant load balancers that inflated the monthly bill, and the team needed to ship features without fear of breaking the platform. At the same time there were unexplained billing spikes —especially in data transfer— and oversized servers left running idle, plus databases fragmented across multiple standalone servers with no high-availability guarantees.
Solution
Craftech executed a full modernization toward a scalable container-based architecture: fully isolated Development, Staging and Production environments, standardized deployment pipelines and capacity autoscaling driven by real demand. On top of that it ran a deep FinOps audit that found and stopped unnecessary network downloads repeating thousands of times per day, applied rightsizing based on historical usage, configured automatic bin-packing every 5 minutes to shut down empty servers, and added spot instances with per-application cost dashboards. Finally it modernized the databases, migrating them to a managed enterprise-grade engine with tailored migration strategies to avoid prolonged downtime.
Results
Network consolidation removed the redundancy inflating the bill: more than 80 load balancers were reduced to one per environment (−98% redundant network resources). Environment separation delivered zero operational risk —developers test without touching production data— and automation accelerated time-to-market. The FinOps audit stopped the capital leak from data transfer, bin-packing and spot instances drive continuous savings, and dashboards gave per-application cost transparency. Database modernization left high availability and resilience in place, with savings of more than US$100/month per optimized server and a migration transparent to end users.
Stack & technologies