Scale VMS to Thousands Cameras per server
Budget: $750 – $1,500 USD
I need help pushing our video-management platform from handling just ten live camera feeds per instance to reliably supporting thousands on AWS. The immediate focus is the Video Management Software itself—profiling, refactoring, and tuning whatever is bottlenecked in our current stack so we can bring feed density up dramatically.
Today the limit appears to be pure software performance, not the hardware or network. I want an engineer who can jump into the existing codebase, identify CPU/GPU hotspots, streamline the video pipeline, and introduce whatever techniques are required—whether that’s smarter thread management, GPU off-loading, micro-services in ECS/EKS, Rust/C++ modules, or leveraging services like MediaConnect and Kinesis Video Streams.
Because our AI analytics run alongside the VMS, every optimization you apply to frame decoding, buffering, or data flow must keep the inference path in mind; no regression there. Once you’ve tuned the core, we’ll benchmark side-by-side on comparable EC2 instances to prove we can scale into the thousands.
Deliverables
• A diagnostics report pinpointing performance choke points
• Refactored or re-configured VMS components that lift camera capacity by an order of magnitude (or more)
• Automated deployment scripts/CloudFormation or Terraform for the new setup
• A concise performance benchmark and step-by-step runbook so my team can replicate, monitor, and maintain the gains
If you thrive on low-level performance work, know AWS inside out, and have dealt with massive live-video workloads before, let’s talk.
Today the limit appears to be pure software performance, not the hardware or network. I want an engineer who can jump into the existing codebase, identify CPU/GPU hotspots, streamline the video pipeline, and introduce whatever techniques are required—whether that’s smarter thread management, GPU off-loading, micro-services in ECS/EKS, Rust/C++ modules, or leveraging services like MediaConnect and Kinesis Video Streams.
Because our AI analytics run alongside the VMS, every optimization you apply to frame decoding, buffering, or data flow must keep the inference path in mind; no regression there. Once you’ve tuned the core, we’ll benchmark side-by-side on comparable EC2 instances to prove we can scale into the thousands.
Deliverables
• A diagnostics report pinpointing performance choke points
• Refactored or re-configured VMS components that lift camera capacity by an order of magnitude (or more)
• Automated deployment scripts/CloudFormation or Terraform for the new setup
• A concise performance benchmark and step-by-step runbook so my team can replicate, monitor, and maintain the gains
If you thrive on low-level performance work, know AWS inside out, and have dealt with massive live-video workloads before, let’s talk.