In-person · Las Vegas · November 30 - December 4 2026

GoogleCloudNext26-hero-BG

#### Beyond fine-tuning: RL for reasoning agents with AWS Trainium

<p class="date-time">DATE & TIME COMING SOON</p>

Supervised fine-tuning enables your model to follow instructions, but teaching it to reason requires reinforcement learning.In this chalk talk, we explore what techniques like RLHF and GRPO bring to the table and how to run them on Trainium with the AWS Neuron software stack.
<a class="blue-btn" style="margin-right:10px;" href="https://registration.awsevents.com/flow/awsevents/reinvent2026/eventcatalog/page/eventcatalog?search=AIM312 ">Save to re:Invent Agenda</a> <i style="font-size:12px;">Space is limited — RSVP to save your spot.</i> 

<style>
  #chalk-talk .hexagon {
     display: none !important;
    }
  .event-sections .hexagon {
     display: none !important;
    }
</style>

[![AWS reInvent-RL](//images.ctfassets.net/xjan103pcp94/5STiFeStAEeYwBYgs8ZI8O/618ba636ac5079f93f637fd464699ce1/reInvent-RL-3x2_1.png)](https://registration.awsevents.com/flow/awsevents/reinvent2026/eventcatalog/page/eventcatalog?search=AIM312)

#### Roundtable: Inference & Post-training

<p class="date-time">MON NOV 30 · 6:30 PM - 9:30 PM</p>

An invite-only dinner designed for AI architects and practitioners to discuss topics around GPU fleet management and utilization, the role of AI in modern data pipelines and infrastructure considerations for the shift to RL. Sponsored by Anyscale, WEKA, Intel and Amazon EKS team. 
<a class="blue-btn" style="margin-right:10px;" href="https://luma.com/anyscale-v6v9 " style="margin-top:0px;">Request Invite</a> <i style="font-size:12px;">Space is limited — RSPV to save your spot.</i>

<style>
  #chalk-talk .hexagon {
     display: none !important;
    }
  .event-sections .hexagon {
     display: none !important;
    }
</style>

[![re invent dinner](//images.ctfassets.net/xjan103pcp94/3vfmReDdyI9yF3U5CzSwaZ/1ee0f04337471603e4efc88ae0409134/re_invent_dinner.png)](https://luma.com/anyscale-v6v9)

GoogleCloudNext26-lets-connect-BG

## Let’s Connect
Grab time with one of our AI experts who can help you plan, build and deploy your next big project with Ray, the de-facto framework used for RL and other AI workloads running at scale. 

AWS re:Invent x Anyscale

AWS reInvent x Anyscale Social Share

Upcoming Webinar: Production-ready Distributed Inference with Ray Serve: 

Powered by Ray, Anyscale empowers AI builders to run and scale all ML and AI workloads on any cloud and on-prem.

Ray on the Road

AWS re:Invent x Anyscale

Chalk talk

In person - AIM312

Beyond fine-tuning: RL for reasoning agents with AWS Trainium

Dinner

Roundtable: Inference & Post-training

Let’s Connect