Recent advancements in Neural Machine Translation (NMT) have resolved the challenges of traditional translation by requiring only a small amount of data for training and can translate over a small number of training words. In this study, we implemented an encoder-decoder architecture using Long Short-Term Memory (LSTM) neural network with attention for translating English to Hindi. While extensive work has been done on high-resource foreign languages, low-resource languages like Hindi lack state-of-the-art preprocessing. Despite this, our model achieved a competitive BLEU score of 32.59. Transformer-based systems trained on large-scale corpora (with millions of sentence pairs) typically achieve BLEU scores around 33–34. Our model, on the other hand, was trained on a moderately sized dataset, the CLARIN English–Hindi corpus, using only about 30,000 sentence pairs. However, our model notably learned word alignments, handled long-range dependencies, and produced fluent translations, effectively managing common NMT challenges such as exposure bias and data sparsity. These results provide confidence on LSTM-attention architectures for low-resource English– Hindi translation tasks.