#vllm

Spark Pulse is a web control plane for spark-vllm-docker

Zero trust LLM gateway. OpenAI-compatible proxy with semantic routing and load balancing across OpenAI, Anthropic, Ollama, vLLM, and any compatible backend. Identity-based access, virtual API keys, and end-to-end encryption via OpenZiti

Manifest-driven 21-stage agent pipeline for Python — 5 LLM backends (Anthropic / OpenAI / Google / vLLM / Claude Code CLI), tools, skills, memory, sandboxed CLI runs, MCP. The engine behind Geny. Apache-2.0.

An out-of-tree vLLM plugin for Mobilint NPU runtime integration.

Build AI agent fleets. 3 lines, any backend, local or cloud. Effort-based model routing, 48 identities, knowledge graph memory, fleet orchestration.

Local-first AI coding agent runtime — your models, fail-closed safety, evidence-gated self-improvement

TurboQuant+ KV cache compression for vLLM. 3.8x smaller KV cache, same conversation quality. Fused CUDA kernels with automatic PyTorch fallback.

OwlCoda — local-first foundation for governed AI business execution: evidence, replaceable executors, admitted results, and explicit authority boundaries.

Model-agnostic agentic runtime for the terminal — any local model becomes a capable agent. The intelligence lives in the harness, not the weights.