Cohort 1

Senior Data Scientist

Fernando Meira

Gain the key skills for designing effective agents and optimizing their performance. Dive deep into evaluations, tools, MCP, and RL.

Will Brown

Kyle Corbitt

Production-Ready Agent Engineering: From MCP to RL

Will Brown & Kyle Corbitt

<h3>Why I enrolled</h3>
My goals are to:<ul class="list-disc ml-8"><li>Explore best-practice MCP implementations, especially those leveraging custom APIs and data.</li><li>Learn how to integrate reinforcement learning into custom tasks and apply it across my current projects and products.</li></ul>


 <h3>A breakthrough moment</h3>Definetely helping me estabilsh my pipeline process: - Define well enough a task and start thinking about evals and ways to evaluate the model - Run benchmarks and optimize prompts as most as possible - Plan and run RL with multiple parameters and testing different hypothesis


<h3>What I’ll apply</h3>
I have learned a lot of valuable things from this course, like: - A clear pipeline on how to start an agent optimization (from simple benchmarks and prompt optimization to then move to RL). - The simple really pays off when getting to do RL: using/or being stuck with many agent frameworks can be a big blocker when trying to apply different reward models on RL pipelines - Iterate fast and learn as much as possible Excite to put in practice: - Try different reward models and iterate more on the problems I am currently working on - Using Will's optional classes to get more intuition into RL algos and try my own experiments - Get to good practical results like the ones demonstrated by Kyle with Agent-E

Certificate

About this course