ApexApex

Formula 1 strategy calls are made in seconds and decide races. From the outside, the reasoning is invisible: tyre degradation, fuel correction, undercut windows, safety-car probability, all of it has to be weighed at once. Existing tools either show raw data or hand you a number without the reasoning behind it.

Apex is built as a workbench, not a prediction box. Physics-based models, machine learning and Monte Carlo simulation meet in one interface, and every recommendation can explain itself.

• Live session tracking via OpenF1, streamed over WebSocket
• Tyre degradation: online RLS estimators plus a neural model, with cliff-risk prediction
• Counterfactual analysis: in-race and pre-race "what if we had" scenarios
• Event-sourced lap processing: stint detection, anomaly filtering, fuel correction
• Explainability engine: every recommendation ships with a sensitivity analysis
• Natural-language race summaries in Turkish and English

498 tests pass and the pipeline runs against live race data. What Apex demonstrates is this: deep domain complexity can be turned into a working, tested system that explains its own reasoning.