Developed a system which is able to generate a summarized text from an environmental image using USB camera.The model works on Resnet - 152 (for image feature extraction) and LSTM (for text generation) achieving the BLEU score of 42.3.

The model is able to summarize a food restaurant review in small tags that could help the restaurant owner and visitors.Model is based on the encoder-decoder Bahdanau Attention based recurrent neural network architecture giving the ROUGE-2 of72.6.

Developed a system that recommends the movies that best matches with the given movie.It involved feature selection and training in an unsupervised manner using KMeans clustering.

Built a web-based system that is currently being used by our University for management and tracking results and staffs in 72 differentdepartments eliminating the paperwork and reducing the delay in the process..

Developed a system using Bi-Directional Attention Flow that would be able to give answers based on the information given. in the article.
The model is trained on SQUAD dataset giving the F1 score of 62% on the test dataset.