The non-destructive characterization of an arbitrary mixture of beta emitters is challenging in environmental monitoring, radiation protection, dismantling, and decommissioning (D&D) operations. Continuous spectra generated by beta decay require deconvolution techniques to extract meaningful information. This paper presents a non-destructive measurement method based on a CAM-PIPS (Continous Air Monitoring Passivated Implanted Planar Silicon) detector coupled with a MLEM (Maximum Likelihood Expectation Maximization) algorithm. The MLEM algorithm automatically analyzes measured spectra using response matrices derived from Monte Carlo simulation. A MCNP6.2 numerical model of the detector was developed and validated by comparing simulated and measured responses from standard beta sources. We assessed the performance of the MLEM algorithm for the deconvolution of 14 C, 36 Cl, 90 Sr, and 137 Cs mixed spectra. Finally, we conducted a sensitivity analysis by adding the response of 60 Co, 133 Ba, 152 Eu and 154 Eu, commonly encountered beta-gamma emitters, to the MLEM matrix. The validated method was employed to characterize beta emitters in nuclear graphite samples. This study confirms the potential of the proposed approach for real-time, non-destructive beta spectrometry applications