Frameworks & middleware

Agent frameworks and federated-learning frameworks.

• OpenAI Agents SDK & Responses API — Multi-tool/multi-agent orchestration, built-in tools, streaming, observability.
• Semantic Kernel — Middleware translating model requests into existing functions/plugins.
• AutoGen — Conversational single- and multi-agent programming framework.
• Flower — Framework-agnostic federated AI: simulation-to-deployment, secure agg, DP, audit, OIDC.
• NVIDIA FLARE — Production-oriented FL SDK: admin, monitoring, access control, confidential computing.
• Substra — Production FL for regulated health/biotech; Kubernetes, auditable logs.
• FATE — Industrial-grade FL with HE/MPC secure computation.
• PySyft — Privacy-preserving remote analysis on data that stays put.
• FedML / TensorOpera — Scalable distributed training/deployment across GPUs, cloud, edge, smartphones.
• TensorFlow Federated — Research framework for custom federated computations.
• OpenFL — Open Federated Learning library (LF AI & Data).
• Flotilla — Research system for large-scale, asynchronous, fault-tolerant federations.
• Context Hub — Andrew Ng open content/skills registry for coding agents.