-
-
Input Form of Current Biostats + Skills along with Target Goal
-
Dashboard Output of Tailored Training Journey
-
Personal Progress with Analytical Insights using MongoDB Atlas
-
Exercise Specific Data from MongoDB Atlas Database
-
Detailed Routine and Challenges for each level along with daily checks.
-
Personal Recordings for VLM Analysis + Training Tutorial Videos
-
Calendar to track Progress by Level
-
Project Architecture
Project Story: Grithub - Your Personal AI Coach
What Inspired Us
The idea behind Grithub was born from the desire to provide a personalized training experience—one that mirrors having a real-life coach who not only guides you to reach the skill level of your ideal sports personality but also pushes you, corrects your form, and interacts with you 1:1. We envisioned a platform that could transform sports training into a literal game-changing experience, offering not just training plans but also motivation, feedback, and an interactive companion to keep you on track.
What We Learned
Building Grithub taught us that we are in the age of exponential growth in AI. The ecosystem of AI agents is vast, and the potential of merging their powers can lead to groundbreaking solutions. However, this journey also highlighted a critical insight: to fully harness AI, you need to be better than the AI itself at orchestrating and guiding it. While AI can provide immense support, it still requires human direction to truly thrive. The challenge is not just in creating intelligent systems but in becoming the conductor of a sophisticated AI orchestra.
How We Built It
Grithub was developed on Amazon Bedrock, leveraging the power of OpenAI to create our unique 10-step training regime. Our frontend was built using HTML, CSS, and Node.js, while our backend is primarily developed in Python with FastAPI handling the backend and frontend integration.
To manage our training regime journey data, we utilized MongoDB Atlas for data warehousing, which also serves as the foundation for generating insightful analysis and user progress reports.
For video and image-based feedback and analysis, we integrated Uform--gen2-dpo, a Vision Language Model hosted on CloudFlare. This model analyzes training videos and images, offering valuable insights to the user. We used Streamlit to facilitate video and image input for VLM analysis.
To add a personal touch, we utilized HumeAI's EVI2 to enable emotionally-intelligent, 1-to-1 conversations between our users and the AI agent. This feature ensures that users not only receive practical advice but also feel engaged and motivated throughout their journey.
Challenges We Faced
While setting up individual modules, AI agents, and services was relatively quick, integrating all components into a cohesive system was a significant challenge. Ensuring smooth communication and interaction between these modules was like solving a complex puzzle.
We also encountered difficulties while running the Cloudflare Uform model, as our system lacked sufficient GPU memory to host the model efficiently. The orchestration of AI agents and managing the complexity of various APIs required not only technical skills but also a strategic approach to bring everything together seamlessly.
Despite these hurdles, overcoming these challenges only made the success of Grithub more rewarding. Our team learned, adapted, and innovated every step of the way, ultimately creating a platform that could redefine sports training experiences.
Accomplishments That We're Proud Of
We are incredibly proud of enabling seamless integration across different AI modules and building a well-rounded agentic AI platform. Successfully making numerous APIs interact with each other in an efficient manner is a significant achievement for our team. Our platform showcases the power of AI orchestration and delivers an immersive, dynamic experience for our users.
What's Next for Grithub
We have ambitious plans for the future of Grithub. Our roadmap includes:
- Leaderboard for Users: Introduce a competitive edge with leaderboards to track and celebrate progress.
- Achievements & Challenges: Add daily and weekly challenges, with achievements to boost engagement.
- Social & Community Features: Build an in-app community where users can share achievements and connect with like-minded individuals.
- Real-Time Notifications: Motivate users with timely reminders and personalized nudges to stay on track.
- Scaling & Market Testing: Expand our scope, scale the platform, launch it to a broader audience, and conduct market testing to refine our offering.
The journey of Grithub is just beginning, and we are excited to continue innovating, enhancing, and delivering a platform that can make a real difference in sports training experiences.


Log in or sign up for Devpost to join the conversation.