🏎️ F1 AI Pit Wall: Predictive Analytics & Agentic Engineering

An end-to-end AI dashboard that combines Machine Learning predictions with real-time F1 telemetry and an LLM-powered Race Engineer.

🌟 Core Features

• 📊 Predictive Timing Engine: A Random Forest model (R²: 0.98) that estimates session durations based on circuit, session type, and start time.
• 📡 Live Telemetry Integration: Real-time track weather and driver intervals fetched via the OpenF1 API.
• 🤖 AI Race Engineer: A specialized agent (Llama 3.3) equipped with domain-specific guardrails, capable of analyzing both historical dataframes and live telemetry.
• 📈 Advanced Analytics: Interactive Plotly visualizations for circuit comparisons, feature importance, and duration distributions.

🛠️ Technical Stack

• Languages: Python
• AI/ML: Scikit-Learn, LangChain, Groq (Llama 3.3-70b)
• Data: Pandas, OpenF1 API
• Frontend: Streamlit, Plotly
• Environment: Dotenv for API security

🏗️ Architecture

The system uses a ReAct Agent framework. When a user asks a question, the agent:

1. Checks the historical dataframe for schedule data.
2. Injects Live Telemetry if the OpenF1 API is active.
3. Uses its Internal Knowledge for historical or driver-specific queries.
4. Operates under strict Domain Guardrails to maintain focus on F1.

🚀 Installation & Setup

1. Clone the repo: git clone https://github.com/yourusername/f1-ai-pitwall.git
2. Install dependencies: pip install -r requirements.txt
3. Create a .env file and add your GROQ_API_KEY.
4. Run the app: streamlit run app.py

---

Developed by Mohammad Ayan
Final Year CSE Student