Devanagari script, used in more than 120 South Asian languages including Hindi, Nepali, and Marathi, poses unique challenges due to complex character structures and ligatures. This work focuses on Offline Nepali Handwritten Word Recognition, involving the creation of a novel handwritten dataset of Nepali words and the implementation of deep learning networks, based on Convolutional Recurrent Neural Networks (CRNN) and attention-based encoder-decoder architectures. This work curates a diverse dataset, addressing limitations of existing Hindi words datasets, such as the repetition of words. The resulting dataset consists of 8558 samples taken from 28 individuals. The Deep Learning models were pre-trained on a generated Synthetic Nepali Fonts Dataset and then fine-tuned on our Nepali handwritten dataset and evaluated with accuracy and Character Error Rate (CER), achieving 81% accuracy and 5% CER. The Depth-wise Separable Convolutions and Squeeze-and-Excite(SE) Blocks were also tested to determine their impact on model size and performance, with the best tradeoff of a 63% reduction in model parameters while degrading the CER and accuracy by only approx. 1% and 2% respectively. The experimental results establish baseline metrics, demonstrating the effectiveness of a dataset with unique words. The results are comparable with the state-of-the-art models trained on IIIT-HW-Dev without lexicon decoding or semantic modules.