Violina Doley

My experience in building with AI spans multiple projects and internships focused on natural language processing, machine learning, and data-driven solutions.

As an AI Research Intern at SuperKalam, I worked on LLM fine-tuning with concepts like reinforcement learning from human feedback (RLHF) and parameter optimization, tailoring models for specialized grading purposes. I also worked on prompt engineering for competitive examination assessments, creating and fine-tuning templates for Large Language Models (LLMs) to evaluate answers in over 10 subject areas.

During my internship at IBM, I designed a Retrieval-Augmented Generation (RAG) chatbot for sales professionals, leveraging LangChain and OpenAI APIs to support efficient order booking, product recommendations, and sales pitch generation. This solution was integrated with PDF and SQL databases for data retrieval, enabling real-time, customer-specific insights via an intuitive WhatsApp interface.

In terms of research, my internships at IIT Hyderabad and IIT Guwahati allowed me to gain hands-on experience in deep learning and domain adaptation techniques. At IIT Hyderabad, I worked on VirConv-L and VirConv-T models for autonomous driving, incorporating attention mechanisms to enhance 3D precision, while my research at IIT Guwahati involved transfer learning on omics datasets to address healthcare biases across ethnic groups.

In addition to my internships, I have developed projects like AI for Predictive Maintenance using LSTM and Autoencoders, achieving high accuracy in predicting and detecting hard disk anomalies. This project combined machine learning with a responsive web interface to demonstrate real-world applications of predictive analytics.

Through these experiences, I have built a solid foundation in both the theoretical and practical aspects of AI, focusing on how to apply it to solve complex, real-world problems.