Post-Training and Deploying Open Source Reasoning Models in Foundry | DEM321

From the creator

Go beyond prompt engineering to build custom reasoning models using reinforcement learning in Microsoft Foundry. We'll walk through how to train and fine-tune a model to improve reasoning quality, deploy it into Foundry, and integrate it into an agent workflow. Designed for developers comfortable with code, this session focuses on real implementation details, covering training loops, evaluation, and deployment patterns that directly impact agent performance. Seating for this session is first-come, first-served. Add it to your schedule to plan your day and arrive early to secure a spot. To learn more, please check out these resources: * https://aka.ms/build26-next-steps * https://aka.ms/build/foundrydiscord 𝗦𝗽𝗲𝗮𝗸𝗲𝗿𝘀: * Chris Lauren * Vijay Aski 𝗦𝗲𝘀𝘀𝗶𝗼𝗻 𝗜𝗻𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻: This is one of many sessions from the Microsoft Build 2026 event. View even more sessions on-demand and learn about Microsoft Build at https://build.microsoft.com DEM321 | English (US) | Working with models Demo | (200) Intermediate #MSBuild Chapters: 0:00 - Opening and technical setup delay 00:00:13 - Introduction to AI agents and demo overview 00:01:44 - Challenges in tool calling order, policy adherence, and latency 00:03:32 - Setup of chatbot scenario using Foundry despite technical hiccups 00:04:30 - Comparison with open source model Quen 314B for cost efficiency 00:05:38 - Introduction of Foundry’s new observability and traceability features for agent sessions 00:06:04 - Automatic capture of mo

Published 2026-06-03 · Watch on YouTube ↗
Availability and playback restrictions are controlled by the original channel.