Collaborating with the team to develop an object detection model for drone-based human detection, utilizing pre -trained models like YOLO v5, YOLO v8, and Fast R-CNN to identify the most accurate model.

•Developed a Search Engine for Human Rights Cases aimed at improving customer satisfaction for lawyers, achieving an accuracy of 95%.•Conducted web scraping on various court websites using Selenium to gather legal data.•Utilized Deep learning techniques to generate summaries and metadata for legal case data with a 97% accuracy rate.•Utilized RAG (Retrieval-Augmented Generation) architecture, leveraging models to deliver contextually rich, accurate responses to user queries… Show more •Developed a Search Engine for Human Rights Cases aimed at improving customer satisfaction for lawyers, achieving an accuracy of 95%.•Conducted web scraping on various court websites using Selenium to gather legal data.•Utilized Deep learning techniques to generate summaries and metadata for legal case data with a 97% accuracy rate.•Utilized RAG (Retrieval-Augmented Generation) architecture, leveraging models to deliver contextually rich, accurate responses to user queries, enhancing system usability and information delivery for legal professionals. Show less

• Client Project: ICICI Bank (Jan 2023 - April 2023)o Built an FAQ Chatbot for ICICI Bank’s iPal system with 85% retrieval accuracy, leveraging a fine-tuned BERT model on GCPo Achieved question paraphrasing accuracy of 90% by using the T5 transformer model, with cosine similarity for matching accuracyo Deployed on GCP using Fast API, with Azure OCR integrated for the data extraction pipeline.• In-House Project:(Aug 2022 – Dec 2022)o Developed a Question Paraphraser, used… Show more • Client Project: ICICI Bank (Jan 2023 - April 2023)o Built an FAQ Chatbot for ICICI Bank’s iPal system with 85% retrieval accuracy, leveraging a fine-tuned BERT model on GCPo Achieved question paraphrasing accuracy of 90% by using the T5 transformer model, with cosine similarity for matching accuracyo Deployed on GCP using Fast API, with Azure OCR integrated for the data extraction pipeline.• In-House Project:(Aug 2022 – Dec 2022)o Developed a Question Paraphraser, used for data augmentation technique in NLP, fine-tuned models like T5, Pegasus, and Rasa, and evaluated model performance with BLEU scores.o Trained a question-answering model using pre-trained models like RoBERTa, DistilBERT, BERT and AllenNLP to identify the most accurate model. Created an FAQ generator and deployed APIs for each solution using Flask. Show less