A key challenge facing display engineers today is how to faithfully reproduce creative intent across a wide range of HDR televisions, each with different peak brightness, color volume and processing limitations. While computationally efficient, the static, curve-based Tone Mapping Operators (TMOs) commonly deployed in consumer displays apply a fixed set of transformations that treat all content uniformly. This paper presents a case study on the design and implementation of a dynamic, learning-based alternative to conventional TMOs. We describe the design and early testing of a lightweight convolutional neural network (CNN) trained to act as a real-time, adaptive tone mapping operator within a simulated HDR television processing pipeline.