Traditionally, analytical and numerical methods have been used to model oil and gas exploration and production (E&P) activities. However, these methods have their own disadvantages to which machine learning (ML) provides a solution. With the help of ML, data-driven models can be created using limited data to gain quality insights. In this study, chiefly two ML algorithms have been used- random forest regression and long short-term memory (LSTM) to create an end-to-end workflow which assists crucial E&P activities. The study is based on open-source data from the Volve oilfield released by Equinor. Continuous learning models have been developed for rate of penetration prediction in well drilling and oil and gas production forecasting, where both subsurface and over-ground parameters influence the results. On the other hand, a sonic log prediction model is created using reference wells and validated using blind data, where the results are influenced by subsurface parameters only. These models perform satisfactorily, with the R2 ranging from 0.54 to 0.85. Despite the apparently low R2 values, the models do a pretty good job in capturing the information and insights that can be gleaned from the data.