Rapid and reliable detection of antimicrobial-resistant (AMR) bacteria is critical in combatting the growing threat of multi-drug resistance. Among all the available detection techniques, surface-enhanced Raman scattering (SERS) has emerged as a powerful tool for detecting a vast range of molecules and microorganisms, with significant potential for identifying bacterial strains. However, current studies using SERS predominantly utilize a drop-cast method, where a bacterial sample is dried with the SERS substrate to produce SERS spectra of bacteria. While effective, this method is time-consuming and less consistent. In this study, we propose a liquid-based SERS method for the rapid identification and differentiation of bacteria strains. Using gold nanoparticles (AuNPs) as the SERS substrate, we focus on detecting and distinguishing Escherichia coli and Pseudomonas aeruginosa in aqueous samples. Our study investigates multiple variables including nanoparticle size (30, 50, and 80 nm), surface charge/capping agent (negative and positive), and analyte suspension media (water and 0.9% NaCl). We found that fresh bacteria samples in 0.9% NaCl mixed with citrate coated AuNPs with the size of 50 nm yielded optimal parameters for bacteria detection. Comparing drop-cast and liquid methods for SERS detection, we determined that the liquid method provided clearer and more consistent SERS spectra, facilitating effective analysis using Principal Component Analysis (PCA). Our method reliably differentiated E. coli bacterial strains and identified molecular features responsible for their differentiation. Overall, this manuscript presents a reproducible, straightforward, and efficient approach for bacteria detection and differentiation using SERS, contributing to strategies aimed at combating the spread of antibiotic-resistant bacteria.