The aim of this thesis is to examine translation quality of English to Bangla language pair for existing models. We used GRU based RNN encoder-decoder with Bahdanau attention mechanism for our translation model. We used SUPara parallel corpus which has 70,861 sentences and added our own Novel Dataset created from Gulliver’s Travel. Our best model gives a BLEU score equal to 48.32. The system is developed using TensorFlow framework in Python 3.0 under Kaggle environment.

‘Her Power’ is a government project where 5-month-long technical training is provided to empower the women of Bangladesh. This was a mixed-research (qualitative and quantitative) where we visited 4 Her Power Centers in 2 subdistricts to collect data for the research. Data transformation method was used in this research and the performance of each training center were calculated based on 5 parameters, namely, quality of the trainer, quality of internet, computer facility, logistics support and student performance.