The rapid electrification of transportation and the global shift toward sustainable energy systems have underscored the critical role of Battery Management Systems (BMS) in ensuring the safety, efficiency, and reliability of lithium-ion batteries in electric vehicles (EVs), renewable energy storage, and portable electronics. This Systematic Literature Review (SLR) provides a comprehensive analysis of deep learning (DL) techniques applied to BMS, focusing on state estimation such as State of Charge [SOC], State of Health [SOH]), fault diagnosis, optimization strategies and their implementation on resource-constrained embedded systems like microcontrollers. Adhering to the PRISMA framework, we systematically reviewed 52 studies published between 2018 and 2024, retrieved from IEEE Xplore and Scopus. The findings demonstrate that DL methods, particularly LSTM and CNN-LSTM, significantly enhance state estimation accuracy, achieving SOC errors as low as 1% MAE and SOH errors of 1%-2% RMSE, leveraging reliable datasets like NASA and CALCE . However, fault diagnosis and optimization remain underexplored, with only 3 and 7 studies respectively, lacking standardized metrics to quantify safety and efficiency improvements.