KubeSlice
Avesha Enterprise for KubeSlice
Service Connectivity Layer for managing fleet of clusters for better application performance
KubeBurst
On demand cloud capacity for datacenters
KubeTally
Multi - cluster chargeback by application and teams
KubeAccess
Service gateway for multi-cloud applications
Smart Scaler
Smart Scaler
Predictive autoscaling based on application behaviors
Smart Karpenter
Predictive autonomous scaling of pods and nodes
Elastic GPU Service
EGS
Single/Multi-cluster and multi-cloud GPU provisioning and management platform
Obliq
Obliq
Obliq adds intelligence and autonomy to Kubernetes
EGS Resources
Explore Resources for Elastic GPU Service
Analyst Reports
Navigating Key Metrics for Growth and Success
Blog
Source for Trends, Tips, and Timely Topics
Documentation
The Blueprint for Mastering Tools and Processes
Customer Case Studies
Success stories from our valued customers and partners
Region
United States
Industry
Biotechnology, genome testing
Optimizing genomic workloads & slashing idle spend with Smart Karpenter
GeneDX accelerates precision-medicine research with event-driven AI/ML pipelines that run on Kubernetes across Oracle Kubernetes Engine (OKE) and Azure Kubernetes Service (AKS). Two research teams share a multi-cloud platform that must scale quickly to meet tight turnaround-time (TAT) requirements for genomic analyses.
Background
| Team | Cloud | Pre-Smart Karpenter Workflow |
| Team A | OKE | Static node group sized for peak. |
| Team B | AKS | Identical pattern; idle nodes during troughs, slow starts during spikes. |
Karpenter was not available on Oracle Cloud Infrastructure (OCI), so the OKE clusters had no concept of just-in-time nodes, also, scaling across the estate was reactive and over-provisioned.
Challenges
- Slow surge response: 5-minute pod queue times during traffic spikes.
- 30 % idle cloud spend: node pools padded to avoid cold starts.
- Manual threshold tuning: DevOps tweaked HPA/VPA every release.
- SLO risk: rising TAT threatened clinical commitments.
Solutions
Smart Karpenter fuses Avesha Smart Scaler with Karpenter to predict pod demand in advance and provision the exact nodes required.
| Capability | Impact at GeneDX |
| Predictive pod scaling: RL models analyze latency, RPS, and service dependencies | Pods launch before a spike; queues disappear. |
| Dynamic node provisioning: predictions drive Karpenter for right-sized nodes | No idle padding; nodes spin up/down in < 60 s. |
| Observation → Optimize rollout | Two-week shadow run before full AI control. |
| Continuous learning | Scaling stays accurate as workloads evolve. |
Implementation Steps
- Helm install in “observe” mode; zero YAML changes.
- One-sprint traffic replay to train baseline models.
- 5 % → 100 % cut-over after three clean deployments.
- Cost guardrails: policy caps daily node-hours; Smart Karpenter throttles non-urgent jobs when budget nears limit.
Results
| KPI | Before | After Smart Karpenter | Delta |
| Average node CPU utilization | 48 % | 82 % | +71 % |
| Idle node-hours / month | 1 900 | 520 | −73 % waste |
| P95 pod queue time | 5 m 10 s | < 45 s | 6.8 × faster |
| SLO violations (job TAT) | 12 / month | 0 | 100 % compliance |
| Cloud compute spend | Baseline | −33 % | Savings fund new research lines |
Customer Voice
“Smart Karpenter makes Karpenter proactive. Nodes appear before the load hits and disappear immediately after, cutting a third of our cloud bill while keeping turnaround times rock-solid.” - Director of Genomic ML Platforms, GeneDX
Conclusion
With Smart Karpenter, GeneDX:
- Achieves predictive, hands-free autoscaling across OKE and AKS.
- Eliminates idle spend while boosting utilization above 80 %.
- Meets stringent diagnostic SLOs without manual tuning.
- Gains a reinforcement-learning foundation for future hybrid-cloud growth.
