AI-Driven Performance Testing Platform
Budget: $10 – $30 CAD
I’m building a platform that uses artificial intelligence to plan, execute, and learn from performance tests. The goal is to move beyond simple scripting and give engineering teams real-time insights and self-tuning recommendations while their applications are under load.
Here’s what I need:
• A modular architecture that lets me plug in existing test runners (JMeter, Gatling, k6, etc.) and stream raw metrics into a unified data layer.
• Machine-learning components that detect anomalies, flag bottlenecks automatically, and predict capacity limits based on historical runs.
• A lightweight dashboard (React, Vue, or similar) that visualises throughput, latency, error rates, and any AI-generated recommendations.
• REST or GraphQL APIs so the engine can be triggered from CI/CD pipelines.
• Container-ready deployment scripts (Docker-Compose or Helm charts) and clear documentation so my team can extend the platform after hand-off.
Acceptance criteria
1. I can start a performance test via API and watch live metrics appear in the UI.
2. When a metric breaches a learned threshold, the system raises an alert and stores the finding for later reports.
3. After a run finishes, the platform outputs a concise summary plus AI-generated optimisation tips in JSON and PDF formats.
4. All components build cleanly in our staging Kubernetes cluster using the provided scripts.
If you’ve previously combined performance engineering with data science, or you have open-source examples of similar work, I’d love to review them and discuss the approach before we kick off.
Here’s what I need:
• A modular architecture that lets me plug in existing test runners (JMeter, Gatling, k6, etc.) and stream raw metrics into a unified data layer.
• Machine-learning components that detect anomalies, flag bottlenecks automatically, and predict capacity limits based on historical runs.
• A lightweight dashboard (React, Vue, or similar) that visualises throughput, latency, error rates, and any AI-generated recommendations.
• REST or GraphQL APIs so the engine can be triggered from CI/CD pipelines.
• Container-ready deployment scripts (Docker-Compose or Helm charts) and clear documentation so my team can extend the platform after hand-off.
Acceptance criteria
1. I can start a performance test via API and watch live metrics appear in the UI.
2. When a metric breaches a learned threshold, the system raises an alert and stores the finding for later reports.
3. After a run finishes, the platform outputs a concise summary plus AI-generated optimisation tips in JSON and PDF formats.
4. All components build cleanly in our staging Kubernetes cluster using the provided scripts.
If you’ve previously combined performance engineering with data science, or you have open-source examples of similar work, I’d love to review them and discuss the approach before we kick off.