Terradev

•
25
•

Orchestrates multi-cloud GPU infrastructure for AI workloads, optimizing provisioning and deployment across various providers.

Terradev is an open-source, multi-cloud GPU infrastructure platform that provides a CLI for imperative AI workload orchestration. It streamlines the entire machine learning lifecycle by automating compute-provisioning, optimizing GPU topology for peak performance, and accelerating deployments 3-5x faster than traditional methods. Key capabilities include intelligent instance selection (spot/on-demand), event-driven triggers, environment promotion with audit trails, automatic lineage tracking for compliance, and specialized optimizations for LLM serving frameworks like SGLang and vLLM, including support for Mixture-of-Experts (MoE) models and disaggregated inference architectures. Terradev ensures cost-effective and high-performance operation across over 20 cloud providers, including bare metal and virtual machine GPU offerings.

주요 기능

01Real-time Cost Analytics and Instance Optimization
02Cross-Cloud GPU Provisioning with NUMA-aware Topology Optimization
03Automated ML Lifecycle Management with Event-Driven Triggers
04Environment Promotion and Automatic Lineage Tracking for MLOps
05Workload-Specific Auto-Optimization for SGLang, vLLM, and MoE Models
0611 GitHub stars

사용 사례

01Rapidly provisioning and optimizing GPU clusters for AI/ML training and inference across multiple clouds.
02Automating production-grade ML pipelines with lifecycle management, triggers, and artifact lineage.
03Deploying and optimizing high-performance LLM inference servers, including MoE models and disaggregated architectures.