PIT
WALL
AI
Analytics
Predictor
Commentary
About
● Live
F1 Intelligence Platform — Season 2024
PIT
WALL
AI
Real telemetry · Machine learning · AI commentary
ENTER DASHBOARD →
Click to enter
VER
LEADER · +0.000
LEC
P2 · +2.455s
NOR
P3 · +7.330s
HAM
P4 · +11.01s
SAI
P5 · +14.88s
DATA
22,126 laps loaded
MAE
2.03s · XGBoost
RACES
15 GPs · 2022–2024
VER
LEADER · +0.000
LEC
P2 · +2.455s
NOR
P3 · +7.330s
HAM
P4 · +11.01s
SAI
P5 · +14.88s
DATA
22,126 laps loaded
MAE
2.03s · XGBoost
RACES
15 GPs · 2022–2024
22,126
Real F1 laps in dataset
2.03s
Model mean absolute error
15
15 GPs · 2022–2024
3
AI commentary styles
// Race Analytics
Live Race
Dashboard
Driver standings — Monaco 2024
01
LECLERC
Ferrari
1:13.842
LEADER
02
PIASTRI
McLaren
1:14.011
+2.455s
03
SAINZ
Ferrari
1:14.203
+4.112s
04
NORRIS
McLaren
1:14.441
+7.330s
05
VERSTAPPEN
Red Bull
1:14.892
+11.01s
06
HAMILTON
Mercedes
1:15.102
+14.22s
Lap time distribution
91.4
s
avg lap
Model performance
98.4%
XGBoost · 22,126 laps · MAE 2.03s
// ML Model
Lap Time
Predictor
Configure prediction
Tyre compound
SOFT
MEDIUM
HARD
Tyre age (laps)
Lap number
Stint
▶ Run Prediction
Predicted lap time
—
Model MAE ±2.03s · XGBoost
Model metrics
Prediction accuracy
98.4%
Training coverage
22,126 laps
Race coverage
6 GPs
Algorithm
XGBoost Regressor
300 estimators · depth 6 · lr 0.05
// AI Commentary
Race
Commentator
Select style
Martin Brundle
David Croft
Data Analyst
Click generate to produce AI race commentary from live lap data...
▶ Generate Commentary
// Architecture
The
Pipeline
01 — Ingest
Data Pipeline
FastF1 API pulls real telemetry, lap times, tyre data from every race weekend. Stored in Firebase Firestore.
02 — Train
ML Model
XGBoost regressor trained on 22,126 laps. Features: tyre age, compound, stint, driver, circuit.
03 — Predict
Inference
Real-time lap time predictions with 2.03s MAE. Bayesian championship simulator runs Monte Carlo outcomes.
04 — Narrate
LLM Layer
Groq LLaMA 70B generates live race commentary in three styles from raw telemetry data.