I’m building an AI/ML solution that will let an SUV sense its surroundings and take over key driving tasks. The initial release must reliably handle lane keeping, execute automatic parking manoeuvres in tight spaces, and recognise pedestrians early enough to trigger safe responses. You’ll design and train the perception and control models, select and fuse sensor data (camera, LiDAR or radar—your call), and deliver code that can run in real-time on an automotive-grade computer. I already have access to a simulation environment and a small-scale test vehicle; your models should slot straight into this pipeline so we can iterate quickly before any road trials. Discrete deliverables • End-to-end model architecture and training scripts • Trained weights with documented performance metrics for the three features above • Integration wrapper that lets me drop the system into ROS or a similar middleware • A brief test report that shows behaviour in simulation and outlines next-step improvements I value clean, well-commented code (Python/C++ preferred) and clear explanations of any hyper-parameter choices or data-augmentation tricks you apply. If you have prior work on autonomous driving modules, feel free to reference it—the more relevant, the better.