AI Infrastructure & MLOps

FLAML provides a fast Python library for automated machine learning and tuning.

Baseten provides an inference platform for serving and scaling open-source and custom AI models in production.

Kuberns uses AI agents to deploy, scale, and manage applications on cloud infrastructure.

GPTCache provides semantic caching for large language models and integrates with LangChain and LlamaIndex.

OneCompression is a Python package for compressing large language models.

Private AI for individuals, teams, and organizations. Workspaces, agents, models, and knowledge—one connected system on your terms.

Kale simplifies data science workflows for deploying Kubeflow Pipelines.

DeepDetect provides a deep learning server and command-line interface for Torch and TensorRT.

Valohai provides MLOps tools for comparing, evaluating, and shipping configurations for AI products using specialized models.

Xinference runs language, speech, and multimodal models through a unified inference API across cloud, on premises, or laptop setups.

Toolhouse turns AI chats into AI workers that can perform tasks while users are away.

oneDNN provides a library of optimized deep neural network operations for oneAPI.

Mosec serves machine learning models with dynamic batching and CPU or GPU pipelines for inference workloads.

OmniRoute routes requests across hundreds of providers through an OpenAI-compatible endpoint with automatic fallback.

Galileo evaluates, monitors, and protects generative AI applications through observability tools for AI teams.

The fuku-ml project provides a machine learning library described as simple to use.

Langfuse provides open-source evaluation and observability tools for tracing and improving AI agents.

ClearML combines experiment and data management, pipelines, scheduling, orchestration, and model serving for AI workloads.

DyNet provides tools for building dynamic neural networks and deep learning models.

Keras provides a deep learning API for building and training machine learning models.

ModelDB tracks machine learning model versions, metadata, and experiments.

Manifest routes language model requests from agents to providers, with subscription, usage-based, and local model options.

Kedro helps create reproducible, maintainable, and modular data science and data engineering pipelines.

Sinabs supports fast training of spiking neural networks with PyTorch and inference on neuromorphic hardware.