Portfolio

Navaneeth Krishnan

AI infrastructure engineer building agentic systems, inference infrastructure, GPU orchestration, and practical evaluation tools.

MCP-native agentsLLM evaluationGPU orchestrationLiteLLM routingFine-tuningComputer visionProduction billing

Systems shipped

Main projects

01MCP-native control plane

GPU infrastructure control plane that lets AI agents provision, schedule, monitor, and route training or inference workloads across providers.

Built an MCP-native infrastructure control plane exposing GPU provisioning, scheduling, job lifecycle management, usage tracking, and billing-aware routing as agent-callable tools.
Designed multi-cloud infrastructure abstraction across RunPod, Lambda Labs, and vast.ai, covering provider selection, workload placement, autoscaling, cold-start optimization, fallback handling, and operational visibility.
Worked on the core reliability problem behind heterogeneous infrastructure: making AI workloads portable, observable, cost-aware, and practical to automate.
Worked directly with B2B customers and prospects to understand workflow constraints, translate requirements into technical systems, and design field-engineering style automation solutions.

MCPPythonFastAPIGPU schedulingMulti-cloudBillingAutoscaling

Research / CV / evaluation