Problem: Build a deep learning model to predict steering angles from road images and control a virtual car in real time.

Steps Taken:
-Collected 17K+ custom driving images in the Udacity simulator
-Preprocessed data (cropping, resizing, color space conversion, normalization) and augmented with flips, brightness, and translations
-Designed and trained a CNN (NVIDIA-inspired) using TensorFlow/Keras
-Deployed real-time inference using Flask and SocketIO

Result:
Successfully deployed a model that autonomously drove a virtual car across multiple maps with smooth, accurate steering predictions in real-time.