Building intelligent agents to handle complex, real-world tasks can be difficult. Additionally, rather than relying solely on large, pre-trained foundation models, organizations often need to fine-tune and adapt smaller, more specific models to perform better in their specific use cases. The AWS AI League offers an innovative program to help enterprises tackle the challenges of building advanced AI capabilities through exciting competitions that foster innovation in agentic AI and model optimization.
In 2025, the first AWS AI League competition attracted the attention of developers, data scientists, and business leaders globally. They come together to solve critical problems using the latest AI tools and technologies. The grand finale of AWS re:Invent 2025 was an exciting showcase of their talent and skills. Cross-functional teams from leading organizations competed head-to-head, demonstrating their ability to generate effective signals, improve models, and build powerful AI agents.
Congratulations to our 2025 AWS AI League Champions! These three extraordinary builders emerged victorious sharing the $25,000 prize pool after intense competition:
- 1st Place: Hemant Vediyara from Cisco
- 2nd place: Ross Williams from Akfar
- Third place: Deepesh Khanna from Capital One
Figure 1: Left to right: Ross, Hemant, Deepesh
This post explains how the AWS AI League program can be used to host AI competitions that can help participants experience model optimization and agent building concepts, apply these to tackle real-world business challenges, and showcase their innovative solutions through engaging, game-style formats. We highlight new agentive AI and model optimization challenges, where enterprises can apply to host internal tournaments using AWS credits, and developers can compete in AWS events.
To get started, visit the AWS AI League product page.
What is AWS AI League Championship?
The AWS AI League experience begins with a 2-hour hands-on workshop led by AWS experts, followed by a self-paced experiment. The journey culminates in an engaging, game show-style grand finale, where you showcase your AI creations and solutions to solve critical business challenges. The following figure shows these three stages.
Figure 2: AWS AI League Championship Stage
Building on the success of the 2025 program, we are excited to announce the launch of the AWS AI League 2026 Championships. This year, the competition features two new challenges that allow participants to really test their AI skills:
- The Agentic AI Challenge allows you to create intelligent agents using Amazon Bedrock AgentCore. Competitors create customized agent architectures to tackle real-world business problems.
- Complementing the Agentic AI Challenge, the Model Optimization Challenge now utilizes the latest fine-tuning recipes in SageMaker Studio. Here you customize models for specific use cases.
For the 2026 AI League Championships, the prize pool doubles to $50,000, offering tracks to developers at various skill levels from beginners to advanced practitioners.
Create intelligent agents with the Agentic AI Challenge
AWS AI League now introduces an exciting agentic AI challenge, where you build intelligent agents using Amazon Bedrock AgentCore to solve complex problems in a dynamic, game-style competition. In this challenge, agents move through a maze-like grid environment and face various challenges while searching for treasure. These challenges map out real-world use cases, testing the agents’ ability to handle inappropriate content, execute code, use the browser, and more.
Agents have a time limit to traverse the map, collect points, and overcome obstacles before reaching the treasure. The more points they earn, the higher their rank on the leaderboard. You can fully customize your agents using Amazon Bedrock AgentCore primitives, enabling you to more securely scale and manage production-grade agents. You can also choose specific models for supervisors and sub-agents, as well as create custom tools like Bedrock Railings, AgentCore memory, and AWS Lambda functions to help your agents tackle challenges. The following diagram shows the obstacles that the agent must overcome while traveling to reach the treasure.
Figure 3: AWS AI League Agentic Challenge
AWS AI League provides a complete user interface (UI) for users to build their own intelligent agent solutions. You can use this no-code UI to build multi-agent architectures and tools, integrate various components like Amazon SageMaker Studio CodeEditor for interactive coding of custom Lambda functions and tools. This allows you to fully develop and customize your agent-based solutions within the AWS AI League website, without the need to leave the environment.
The following screenshots demonstrate the agent creation experience within the AWS AI League website.
Figure 4: AWS AI League Agent tool
Figure 5: AWS AI League Multi Agent Architecture
Throughout the competition, users receive real-time agent performance feedback, with a large language model (LLM) evaluator providing evaluation to aid iteration. The following image shows how the agent is evaluated during challenges.
Figure 6: AWS AI League Agent Challenge Evaluation
In the grand finale, the top finalists take the stage to showcase the capabilities of their agents in a live, game-show format, demonstrating the power and versatility of agentic AI in solving complex, multi-step problems. Evaluation criteria include time efficiency, accuracy in solving challenges, agent planning, and token consumption efficiency. The following snapshot shows the final round of the Grand Finale at Re:Invent 2025.
Figure 7: AWS AI League Re:Invent 2025 Grand Finale
Optimize models to outperform larger models
The AWS AI League is expanding the scope of its Model Optimization Challenge, allowing you to use the latest advances in fine-tuning techniques.
You can access the new model optimization experience within Amazon SageMaker Studio, where you can use powerful new training recipes. The goal is to develop highly effective, domain-specific models that can outperform the performance of larger, reference models.
The challenge begins with honing your model optimization skills. Using the tools and techniques you learn, you apply advanced fine-tuning methods to help increase the performance of your model. After your models are optimized, the true test begins. The model is presented on a leaderboard for performance evaluation against a reference model. The model earns points whenever the automated judge judges your optimized model’s response to be more accurate and comprehensive than the reference model’s output. You can showcase your advanced skills, rise to the top of the leaderboard and potentially unlock new opportunities for your organizations.
During the challenge, you receive real-time feedback on your model’s performance from an automated evaluator when you submit to the leaderboard. The leaderboard evaluates submissions against a reference dataset throughout the competition, providing immediate feedback on accuracy to help you iterate and improve your solutions. The following image shows how AI critique is used to evaluate optimized models.
Figure 8: AWS AI League Model Optimization Evaluation
In the grand finale, the top finalists showcase the capabilities of their models in a live, game-show format while showcasing their accelerated engineering abilities. During the game show, scoring involves expert evaluation where domain experts and the live audience participate in real-time voting to determine which AI solution best solves real business challenges. The following image shows a quick engineering view of the participant during the grand finale.
Figure 9: AWS AI League Model Optimization Grand Finale Participant View
conclusion
In this post, we explore the new AWS AI League Challenges and how they are changing how organizations approach AI development. At AWS, we’ve learned that the fastest way to drive innovation is competition. With AWS AI League, builders can now showcase their AI skills, compete, and unlock innovation.
Visit AWS AI League to learn more about hosting an AWS AI League within your organization and learn in-depth about building intelligent agents and optimizing AI models. AWS AI Training Catalog But AWS Skills Builder,
About the authors
mark carp is an ML Architect with the Amazon Sagemaker Services team. He focuses on helping customers design, deploy, and manage ML workloads at large scale. In his free time, he likes to travel and discover new places.
Natasya K. Idris Product Marketing Manager for AWS AI/ML Gamified Learning programs. She is passionate about democratizing AI/ML skills through engaging and practical educational initiatives that bridge the gap between advanced technology and practical business implementation. His expertise in building learning communities and advancing digital innovation is shaping his approach to creating impactful AI education programs. Outside of work, Natasya enjoys traveling, cooking Southeast Asian cuisine, and nature walks.
