Cut AWS costs by 73% while achieving 100% uptime during critical exam periods

Company: AtomsLearning

Industry: Medical Education / EdTech

Location: India

Challenge: Medical exam preparation platform experiencing severe traffic fluctuations with rigid infrastructure unable to scale elastically, resulting in performance degradation during peak exam periods and wasteful spending during quiet cycles.

AtomsLearning faced dramatic traffic fluctuations inherent to exam preparation platforms, with AWS costs at $5,200/month. The static EC2 infrastructure couldn't adapt to cyclical usage patterns.

Infrastructure Pain Points:

• Peak Traffic Overwhelm — MockExam microservices experienced severe performance deterioration during unpredictable exam night traffic surges
• Infrastructure Rigidity — Static EC2 deployments couldn't scale elastically, leading to resource over-provisioning during quiet cycles
• Cost Inefficiency — Fixed infrastructure costs persisted year-round despite seasonal usage variations

Original Architecture Limitations:

• Fixed capacity allocation regardless of demand
• Reactive scaling responses with significant delays
• Monolithic deployment restricting component-level scaling independence

I identified two distinct operational modes requiring fundamentally different infrastructure strategies:

• Exam Season Mode: High traffic volumes with unpredictable surges requiring maximum elasticity and performance capability
• Preparation Cycles: Steady, predictable usage patterns requiring cost optimization and operational efficiency

The fundamental insight: Rather than designing one system for all scenarios, build two complementary systems that automatically transition based on scheduling and demand patterns.

Phase 1: Amazon EKS for Peak Performance

• Containerized microservices enabling granular scaling at the service level
• Horizontal Pod Autoscaling based on CPU, memory, and custom application metrics
• CI/CD integration with seamless deployment pipelines and rollback capabilities

Phase 2: K3s on EC2 for Cost Efficiency

• Lightweight Kubernetes reducing operational overhead by 60%
• ARM Graviton2 instances offering 20% superior price-performance ratio
• Demand-aligned scaling policies precisely matching resources to actual consumption

Implementation Timeline: 7 Days

Complete ownership of AWS and Kubernetes infrastructure, enabling the team to concentrate on product development rather than infrastructure emergencies.

Additional Benefits:

• Elastic scaling adapting automatically to traffic patterns without manual intervention
• Streamlined operations during non-peak periods with reduced complexity
• Confidence to scale user base without infrastructure constraints
• Complete infrastructure ownership enabling focus on product development