The increasing volume of waste generated by various activities has increased interest in using waste to create sustainable construction materials to achieve possible benefits. In addition, using recycled materials to produce fresh concrete is a desirable option because of its low cost, lower landfill space requirement, and the completed concrete quality. Therefore, an experimental inquiry is undertaken to ascertain the impacts of up to 20 wt% cement displaced by Volcanic Pumice Powder (VPP) with the incorporation of 1% and 2% Recycled Nylon Fiber (RNF) on the mechanical properties of concrete composites following room temperature to high-temperature (600 °C) exposure. Fresh concrete characteristics tests were performed, including slump, compacting factor, Kelly ball penetration, and density. The heat resistance of the concrete was then measured by calculating the percentage decrease in weight, the splitting tensile strength, and the compressive strength of the specimens. Heating mainly raised VPP's pozzolanic reactivity and lowered high vapor pressure through melting RNF. Therefore, VPP and RNF-treated concrete had superior mechanical performance than control concrete even when exposed to elevated temperatures. Further, the microstructural modifications brought on by RNF and VPP additions were also explored by deploying Scanning Electron Microscopy (SEM). The use of VPP in concrete led to an improvement in fresh properties, while RNF demonstrated deterioration in the same qualities. Despite this, supervised machine learning techniques are a central focus of this investigation because of their potential to predict concrete characteristics accurately. To predict the fresh and mechanical characteristics of concrete, both the Random Forest (RF) and the K-Nearest Neighbors (KNN) algorithm, along with their ensemble model counterparts, were explored. The outcomes revealed that RNF and VPP considerably improved the concrete's heat resilience and mechanical characteristics and halted the concrete composites' explosive spalling behavior at 600 °C temperatures. To prevent strength loss at high temperatures, it was discovered that adding 1% RNF content to concrete with 10% VPP was the best combination. In addition, the high coefficient correlation (R2) value of the RF and the ensemble model indicates great accuracy in outcome prediction, while the low R2 value of the KNN model indicates that the KNN model is less accurate.