D Ruchitha Email & Phone Number

Bengaluru, Ka, In

Enhancing NLP Models: Fine-Tuning GPT - 3.5 Turbo and Llama2-7B-Chat To Optimize the performance of GPT-3.5 Turbo and Llama2-7B-Chat models in natural language processing tasks, with a specific focus on improving text generation capabilities.Led fine-tuning initiatives for GPT-3.5 Turbo and Llama2-7B-Chat models, enhancing their NLP capabilities, particularly in text generation and question-answering.Utilized ML techniques to adapt models for specific NLP challenges, emphasizing large language model utilization.Leveraged an iterative approach to fine-tuning, incorporating feedback from human evaluators to continuously enhance model performance and address emerging challenges.Achieved significant improvements:GPT-3.5 Turbo: Trained for 12 epochs, achieving 50.55% human evaluation score.Llama2-7B-Chat: Conducted three trials, achieving up to 80.00% human evaluation score.

Bengaluru, Karnataka, India

Demand Sensing To develop an advanced forecasting and sensing platform capable of accurately sensing demand across various time intervals, including monthly, weekly, daily, and intraday cadences. Additionally, it aims to create a simulator within the platform to enable users to explore different scenarios and make informed decisions based on real-time data insights. Led data management efforts for the Demand Sensing Application, overseeing the curation and maintenance of datasets including The Conference Board Data, Weather Data for Cities, Twitter Keyword lists, CPI data, FRED data, and more.Developed a comprehensive R Markdown Data Procurement notebook to ensure data consistency and understanding across the project team.Designed and implemented HTML and CSS components for the user interface, enhancing the project's visual appeal and usability.Authored Cypress test cases to rigorously test the application's functionality.