RaceSense AI

• 

  Snippet of Dashboard/Driver Overview tab

• 

  Snippet of Lap time Prediction Tab

• 

  Chart showing the range of prediction and actual predicted point

• 

  Snippet of Driving insights Tab

• 

  A section showing Improvements potential chart

• 

  A section showing some actionable recommendations

• 

  Performance pattern Page taken from the deployed streamlit page

• 

  Snippet of Sector Analysis tab

• 

  Snippet of a section from the sector analysis page

• 

  Snippet of Leaderboard section (with zoomed chart, a chart tool function)

• 

  A driver ranking section from the leaderboard page

Inspiration

I recently graduated with degrees in Computer Science and Cybersecurity, and I’ve been looking for a project that would challenge me and excite me. When I discovered the Toyota GR Cup datasets, something clicked instantly; the complexity, the precision, and the engineering behind racing data drew me in. What started as curiosity quickly turned into a full commitment to build something meaningful.

What it does

RaceSense AI is a complete race analytics and driver insight platform built for the Toyota GR Cup. It transforms raw data into predictions, performance analysis, telemetry insights, sector comparisons, and intelligent commentary.

It helps drivers understand their strengths, areas of improvement, and potential lap time gains (amidst other functionalities). The dashboard explains predictions and insights in simple language, allowing anyone to explore race data like a professional race engineer.

How I Built It

I built the entire project solo and from scratch during this hackathon. I used the Indianapolis dataset, combining lap times, sector data, telemetry, weather, and race results.

The backend performs data cleaning, feature creation, and model training. The dashboard is built with Streamlit and Plotly using a fully custom dark-themed UI system. I created:

• A multi-model prediction system
• A sector analysis engine
• A rule-based and AI-enhanced insight generator
• A custom UI framework to maintain visual consistency

Every chart, layout, and interaction was manually designed to give a modern software feel.

To ensure the insights were credible, I studied some real race engineering material and professional analysis techniques, allowing me to interpret the dataset with an expert-level perspective.

Challenges I Ran Into

Many parts of the project pushed me harder than expected. The AI commentary and driver insights were especially challenging because I wanted them to feel intelligent and not repetitive. I had to design new metrics like the Grip Utilization Score to validate the commentary. The Indianapolis dataset also required careful cleaning, merging, and alignment across multiple files. I kept improving the system even after it was functional because I wanted the final version to feel truly complete and useful.

Accomplishments I Am Proud Of

I am proud of how far this project came within the hackathon period. The dashboard is fully functional, visually polished, and deeply analytical. The insights system feels meaningful, the predictive models are stable, and the UI is consistent across all sections. Most of all, I am proud of the discipline it took to push beyond the first working version and refine the project into something I can present to real engineers and drivers.

What I Learned

I learned how to think like a race engineer and how to blend data science with performance storytelling. I gained experience designing a large multi-section dashboard, improving UI consistency across tabs, and solving complex plotting and layout challenges. The experience confirmed the value of iteration; each refinement made the tool more valuable and pushed me to understand the data and user experience more deeply.

What is next for RaceSense AI

The next steps are expanding the dataset to include more races, adding full mobile responsiveness, improving the AI insight engine, and making the dashboard more interactive. I also plan to enable cross-race analysis, allowing drivers to track their performance and make insightful, data-backed improvements across an entire season. I want RaceSense AI to grow into a complete tool for drivers, teams, fans, and analysts.

Built With

• css
• matplotlib
• numpy
• pandas
• plotly
• python
• scikit-learn
• statsmodels
• streamlit