---
title: Race Telemetry
emoji: 🏁
colorFrom: green
colorTo: blue
sdk: docker
app_port: 8501
tags:
- streamlit
- mlops
- motorsport
- telemetry
- machine-learning
pinned: false
short_description: Real-time motorsport telemetry ML  — Pit Wall Dashboard
---

# 🏁 Race Telemetry — Pit Wall Dashboard

> **This is a lightweight UI demo.** The full production MLOps platform is on GitHub → **[nasim-raj-laskar/Race-Telemetry](https://github.com/nasim-raj-laskar/Race-Telemetry/tree/main)**

Streams pre-recorded race telemetry and runs three ML models in real-time:-

| Model | Type | Predicts |
|---|---|---|
| Lap Time Predictor | XGBoost Regression | Lap time (seconds) |
| Gear Optimizer | Random Forest | Recommended gear |
| Driving Behavior | K-Means Clustering | Aggressive vs Smooth |

## Full System Architecture

![Architecture](assets/worklfow.png)

The real system includes FastAPI inference, PostgreSQL, MongoDB Atlas training pipeline, MLflow + DagsHub experiment tracking, PSI drift detection, Prometheus + Grafana observability, and CI/CD via GitHub Actions → AWS ECR → EC2.

## Demo vs Production

| | This Demo | Full Platform |
|---|---|---|
| Models | Bundled `.pkl` | Dynamically loaded from DagsHub |
| Data | Static CSV | PostgreSQL live fetch |
| Inference | Direct in Streamlit | FastAPI backend |
| Monitoring | None | Prometheus + Grafana |
| Deployment | HF Spaces | AWS ECR + EC2 |