ApexLens

🚗 About the Project: ApexLens

💡 Inspiration Formula 1 is the pinnacle of human and machine performance — but behind every lap lies an ocean of raw telemetry data that’s rarely explored beyond simple dashboards. Teams can monitor live data, but finding why something happened — like when a driver managed a perfect overtake in the rain or lost pace during high tire wear — requires manually combing through thousands of data points. We wanted to change that. That’s what inspired ApexLens — an AI-powered telemetry intelligence platform that transforms raw F1 telemetry into relevant and actionable insights.

⚙️ How We Built It We built ApexLens using a Python + FastAPI backend, connected to Pinecone Database for storing and retrieving embeddings of telemetry data. The Gemini API acted as our reasoning engine — interpreting race scenarios, identifying tire degradation patterns, and generating natural-language insights like “Driver A is likely to undercut Driver B due to a 0.8s quicker out-lap on fresher mediums.” On the frontend, we designed an interactive React.js interface powered by JavaScript, allowing users to see actionable insights as the race is going on.

🧠 What We Learned Building ApexLens taught us how to bridge the gap between raw vectorized data and high-level human reasoning. We learned how vector search can contextualize telemetry events beyond numerical stats, how FastAPI can efficiently orchestrate model and database calls, and how LLMs like Gemini can reason through racing logic to produce actionable insights. We also gained deep experience in designing data pipelines that clean, normalize, and embed time-series telemetry into a semantic layer.

🔥 Challenges We Faced Our biggest challenge was aligning vector embeddings with real-world racing semantics. For example, teaching the system that a “tire cliff” isn’t just a drop in lap times, but a compound interaction between track temperature, stint length, and driver style. We also had to optimize query normalization to prevent inconsistent Gemini outputs and ensure real-time performance despite large telemetry datasets. Frontend integration was another hurdle — making AI-generated strategy insights appear fluidly alongside telemetry info required precise API timing and error handling.

🏁 Outcome By the end, ApexLens evolved into more than just a race data tool — it became an AI race strategist, capable of summarizing entire race phases and uncovering hidden narratives within data. It’s a glimpse into how AI can change the way teams, analysts, and fans understand motorsport.