This paper presents a microwave filter design methodology that incorporates physical aspects consistent with fabrication and packaging earlier in the design process for more precise, efficient and cost-effective designs that better mirror the physical reality. The unconventional filter synthesis approach presented makes use of anticipated parasitic effects and lumped impedances that emerge from the physical structure due to the fabrication and packaging for the filter imposed on the designer. In-line filter layouts, positive coupling designs which may be preferable to negative coupling designs, and extracted pole (EP) filters topologies which may be more desirable than cross-coupled resonator designs are theoretically considered and assessed with several realized practical implementations. This paper presents several approaches to improve performance of GaN power amplifiers (PAs) for sub-6 GHz 5G, 6G and beyond wireless communications applications. Various Doherty amplifier design refinements such as frequency-dependent compensating circuit (FDCC) implementations for wideband applications and dual-mode biasing circuits for multiband applications are explored to enhance backoff efficiency, drain efficiency (DE), power-added efficiency (PAE) and gain flatness over frequency. AI-based load and supply modulation optimization and deep neural network configuration of digital predistortion (DPD) implementations to linearize in real time have significant potential for performance enhancement in the future.