Train, Host, and Infer Reasoning Models on Microsoft Foundry
Date:
Breakout Session at Microsoft Ignite 2025, San Francisco, CA
Breakout session (BRK210) at Microsoft Ignite 2025, co-presented with Microsoft’s Azure AI Platform team. The talk walks through the full lifecycle of reasoning models on Microsoft Foundry: training custom reasoning models with reinforcement learning, hosting them for low-latency inference, and shipping them to production with Azure Machine Learning — taking reasoning models from prototype to enterprise deployment for Fortune 500 workloads.