The random blood glucose monitoring is a quick testing at any time to detect hyperglycemia and decrease the risk of common health complications. This paper focuses on a non-invasive in vivo random blood glucose estimation using the Photoacoustic Spectroscopy (PAS) technique. A portable system has been designed to obtain the pre-processed Photoacoustic (PA) signal from a finger using a Laser source of wavelength 905 nm. The Partial Least Square Regression algorithm is implemented where the PA signal feature values are utilized for calibrating coefficients, while the Body Mass Index parameter is included for better accuracy. The study has been conducted on 32 subjects, and the performance is observed with the Clarke Error Grid and Bland-Altman Plots. The Mean Absolute Relative Difference, Mean Absolute Difference, and Root Mean Square values are estimated as 8.9%, 13.75 mg/dl, and 16.13 mg/dl, respectively. Overall, the proposed non-invasive random blood glucose monitoring technique can facilitate the growth of point-of-care applications for frequent monitoring, particularly in resource-limited settings, which facilitates proper diabetic management.