Industries that engage in laser-based material processing are increasingly turning to fiber lasers as one of the most effective laser systems available. Using a pulsed fiber laser system, an experimental investigation of underwater laser micro-channelling on hard to machine thick PMMA has been conducted in this research work. Submerged laser transmission cutting which is also known as underwater laser induced backside machining helps to generate clean kerf edge without presence of mushy region, which is very common during laser machining of thick PMMA with wavelength of near infrared region. Submerged conditions are used here for laser transmission micromachining in order to reduce adverse thermal effects and microcracking or charring inside the material. Pulse frequency, working power, and cutting speed have been selected as the input parameters. As machining responses, the depth of the micro-channel, the width of the kerf, and the width of the heat affected zone (HAZ) have been considered. To find the ideal parametric condition for simultaneously achieving numerous goals, Response Surface Methodology (RSM) and AI-based Teaching Learning-Based Optimisation (TLBO) have been used. The TLBO technique determined that the optimal laser machining settings are a power setting of 13.98 %, a pulse frequency of 50 kHz, and a cutting speed of 0.20 mm/sec. These parameters result in a minimum HAZ width of 4.1874 µm, a minimum kerf width of 22.3698 µm, and a maximum depth of cut of 37.1478 µm. Underwater laser processing reduces heat affected zone and redeposition surrounding micro-channels, creating a cleaner, finer structure.