- Utilized Langchain for automating the generation of custom SQL queries for extracting the required data from SQL databases upon user query.- Also employed Langchain for the creation of embeddings for pdf documents, enabling the extraction of relevant information with higher accuracy and efficiency.

Joined LTIMindtree( Erstwhile Mindtree) on September 30, 2020 as a fresher. Completed the Orchard training in Digital AI (Artificial Intelligence) Track. After compIeting the training I have worked as an Associate Data Scientist in Data Science & Engineering Department and below is a brief summary of my role:- Trained and fine-tuned a faster-rcnn object detection model for detecting different data components such as date, subtotal, total, etc from an invoice. By fine-tuning the model the accuracy of the model increased from 80% to 95%.- Developed the python code for getting the final output by reading the text from the detected data components of the invoice by using the tesseract ocr. Also wrote the python code for calculating the accuracy by comparing the final result with the groundtruth.- Worked with different azure cognitive services like speech to text, vision, etc. Also have implemented the azure blob trigger function.- Developed and implemented search functionality using Azure Cognitive Search for extracting relevant information from a document for the user question. Also made use of openai gpt 3.5 for getting the final answer to the user question from the relevant information extracted earlier.