Constructing clever brokers to deal with advanced, real-world duties might be daunting. Moreover, quite than relying solely on massive, pre-trained basis fashions, organizations usually must fine-tune and customise smaller, extra specialised fashions to outperform them for his or her particular use circumstances. The AWS AI League supplies an modern program to assist enterprises overcome the challenges of constructing superior AI capabilities via thrilling competitions that drive innovation in agentic AI and mannequin customization.
In 2025, the primary AWS AI League competitors captured the eye of builders, information scientists, and enterprise leaders globally. They got here collectively to resolve urgent issues utilizing the most recent AI instruments and strategies. The grand finale at AWS re:Invent 2025 was an thrilling showcase of their ingenuity and expertise. Cross-functional groups from main organizations competed head-to-head, demonstrating their means to craft efficient prompts, fine-tune fashions, and construct highly effective AI brokers.
Congratulations to our 2025 AWS AI League Champions! After intense competitors amongst these three distinctive builders emerged victorious, sharing a $25,000 prize pool:
- 1st Place: Hemanth Vediyera from Cisco
- 2nd Place: Ross Williams from Aqfer
- third Place: Deepesh Khanna from Capital One
Determine 1: Left to proper: Ross, Hemanth, Deepesh
This put up explores how the AWS AI League program can be utilized to host AI competitions that may assist individuals expertise mannequin customization and agent constructing ideas, apply these to sort out real-world enterprise challenges, and showcase their modern options via partaking, game-style codecs. We spotlight the brand new agentic AI and mannequin customization challenges, the place enterprises can apply to host inside tournaments utilizing AWS credit, and builders can compete at AWS occasions.
To get began, go to the AWS AI League product web page.
What’s the AWS AI League Championship?
The AWS AI League expertise begins with a hands-on, 2-hour workshop led by AWS specialists, adopted by self-paced experimentation. The journey culminates in a fascinating, gameshow-style grand finale, the place you showcase your AI creations and options to handle urgent enterprise challenges. The next determine reveals these three steps.
Determine 2: AWS AI League Championship steps
Constructing on the success of the 2025 program, we’re excited to announce the launch of the AWS AI League 2026 Championship. This 12 months, the competitors options two new challenges that permit individuals to actually put their AI expertise to the check:
- The agentic AI Problem permits you to construct clever brokers utilizing Amazon Bedrock AgentCore. Rivals craft personalized agent architectures to sort out real-world enterprise issues.
- Complementing the agentic AI Problem, the mannequin customization Problem now makes use of the most recent fine-tuning recipes in SageMaker Studio. Right here you customise fashions for particular use circumstances.
For the 2026 AI League championship, the prize pool doubles to $50,000, with tracks catering to builders at totally different talent ranges – from rookies to superior practitioners.
Construct clever brokers with the agentic AI problem
The AWS AI League now options an thrilling agentic AI problem, the place you construct clever brokers utilizing Amazon Bedrock AgentCore to resolve advanced issues in a dynamic, game-style competitors. On this problem, brokers navigate via a maze-like grid setting, encountering varied challenges whereas looking for a treasure chest. These challenges map to real-world use circumstances, testing the brokers’ means to deal with inappropriate content material, execute code, use a browser, and extra.
Brokers have a time restrict to traverse the map, gather factors, and overcome the obstacles earlier than reaching the treasure chest. The extra factors they earn, the upper they rank on the leaderboard. You may absolutely customise your brokers utilizing Amazon Bedrock AgentCore primitives, which lets you extra securely scale and handle production-grade brokers. You too can choose particular fashions for supervisor and sub-agents, in addition to create customized instruments reminiscent of Bedrock Guardrails, AgentCore Reminiscence, and AWS Lambda capabilities to assist your brokers navigate the challenges. The next determine depicts the obstacles the agent should overcome whereas touring to achieve the treasure chest.
Determine 3: AWS AI League Agentic Problem
AWS AI League supplies a full person interface (UI) for customers to construct their clever agent options. You need to use this no-code UI to assemble multi-agent architectures and instruments, integrating varied elements reminiscent of Amazon SageMaker Studio CodeEditor for interactive coding of customized Lambda capabilities and instruments. This lets you absolutely develop and customise your agent-based options throughout the AWS AI League web site, without having to go away the setting.
The next screenshots showcase the agent constructing expertise all throughout the AWS AI League web site.
Determine 4: AWS AI League agent instruments
Determine 5: AWS AI League multi agent structure
All through the competitors, customers obtain real-time agent efficiency suggestions, with a big language mannequin (LLM) evaluator offering evaluation to assist with iteration. The next picture showcases how the agent is evaluated throughout challenges.
Determine 6: AWS AI League agent problem analysis
On the grand finale, the highest finalists take the stage to showcase their brokers’ capabilities in a dwell, game-show format, demonstrating the ability and flexibility of agentic AI in fixing advanced, multi-step issues. The analysis standards embrace time effectivity, accuracy in fixing challenges, agent planning, and token consumption effectivity. The next snapshot reveals the ultimate spherical of the Grand Finale at re:Invent 2025.
Determine 7: AWS AI League re:Invent 2025 Grand Finale
Customise fashions to outperform bigger fashions
AWS AI League is increasing the scope of its mannequin customization problem, permitting you to make use of the most recent developments in fine-tuning strategies.
You may entry the brand new mannequin customization expertise inside Amazon SageMaker Studio, the place you need to use highly effective new coaching recipes. The objective is to develop extremely efficient, domain-specific fashions that may outperform the efficiency of bigger, reference fashions.
The problem begins with you honing in in your mannequin customization expertise. Utilizing the instruments and strategies you’ve gotten discovered, you apply superior fine-tuning strategies to assist improve your mannequin’s efficiency. After your fashions are personalized, the true check begins. The fashions are submitted to a leaderboard for efficiency evaluation in opposition to a reference mannequin. The mannequin earns factors every time the automated decide deems your personalized mannequin’s response to be extra correct and complete than the reference mannequin’s output. You may showcase your superior expertise, rise to the highest of the leaderboard, and doubtlessly unlock new alternatives to your organizations.
Through the problem, you obtain real-time suggestions in your mannequin’s efficiency from an automatic evaluator once you undergo the leaderboard. The leaderboard evaluates submissions in opposition to a reference dataset all through the competitors, offering speedy suggestions on accuracy that will help you iterate and enhance your options. The next picture showcases how an AI critique is used to guage the personalized mannequin.
Determine 8: AWS AI League mannequin customization analysis
On the grand finale, the highest finalists reveal their fashions’ capabilities in a dwell, game-show format, showcasing their immediate engineering talents. Through the gameshow, the scoring contains knowledgeable analysis the place area specialists and a dwell viewers take part in real-time voting to find out which AI options greatest clear up actual enterprise challenges. The next picture showcases the participant immediate engineering view throughout a Grand Finale.
Determine 9: AWS AI League mannequin customization Grand Finale participant view
Conclusion
On this put up, we explored the brand new AWS AI League challenges and the way they’re reworking how organizations method AI growth. At AWS, we’ve discovered that the quickest approach to spark innovation is thru competitors. With AWS AI League, builders can now showcase their AI expertise, compete and unlock innovation.
To be taught extra about internet hosting an AWS AI League inside your group go to the AWS AI League and to dive deeper into constructing clever brokers and customizing AI fashions discover AWS AI coaching catalog on AWS Ability Builder.
Concerning the authors
Marc Karp is an ML Architect with the Amazon SageMaker Service staff. He focuses on serving to clients design, deploy, and handle ML workloads at scale. In his spare time, he enjoys touring and exploring new locations.
Natasya Ok. Idries is the Product Advertising Supervisor for AWS AI/ML Gamified Studying Applications. She is obsessed with democratizing AI/ML expertise via partaking and hands-on instructional initiatives that bridge the hole between superior know-how and sensible enterprise implementation. Her experience in constructing studying communities and driving digital innovation continues to form her method to creating impactful AI education schemes. Outdoors of labor, Natasya enjoys touring, cooking Southeast Asian cuisines and exploring nature trails.
