kevin.ai

Inspiration

We were all looking for a reason to start a coding project based on AI and this was the perfect opportunity. Everyone in the group has played copious amounts of Valorant, with our dear team member Kevin even having tournament experience on the Rutgers New Brunswick Val team. It was only fit to name our chatbot after him.

What it does

kevin.ai allows users to build a Valorant esports team based on specific criteria like player region, agent preference, and team roles. It can explain the reasons for its decisions and presents ideas in a clear and organized format.

How we built it

AWS services including Amazon Bedrock make up the cloud infrastructure, and VLR.gg along with an SQL database to store player data. An Amazon Bedrock agent translates user messages into SQL queries to find specific information.

Challenges we ran into

Getting accustomed to AWS and switching our frontend from OpenwebUI to Streamlit were the hardest obstacles.

Accomplishments that we're proud of

While it was difficult at first to determine what criteria to consider, our custom Evaluation Score encompasses VLR rating, agent pool flexibility, and tournament experience into one number for each player. We believe this custom scoring system differentiates us from other projects and helps our chatbot create more well-rounded teams, with accompanying visuals to compare previously made teams.

What we learned

It makes sense that seemingly every tech company is moving toward AI. Especially after some tweaking, the capabilities of our chatbot far exceed our original expectations.

What's next for kevin.ai

We think it would be a fun idea to escalate kevin.ai into a Valorant Esports Fantasy app, where users can draft Valorant teams and compete with others in a league.

WEBSITE INFO

Website: www.kevinai.us

Built With

• amazon-web-services
• bash
• claude
• docker
• javascript
• openwebui
• python
• sqlite
• vlr.gg