Mehedi Hasan

I am Mehedi Hasan, I worked as a Trainee Data Scientist at IIMI and Business Analyst at Turing. I am currently doing a masters in theoretical physics. My research interest is in the intersection of AI and Physics.

I have worked as a Trainee Data Scientist at Intelligent Image Management Ltd., where I worked on large language models employed to extract data from old documents. I was involved in full-stack machine learning engineering, from dataset preparations to LLM training, prediction, and taking user feedback to refine model performance for continuous improvement.  This use case of large language models is unique and has immense challenges. By integrating OCR and LLMs, we improved the accuracy and efficiency of data extraction from old documents.
I also worked on computer vision to detect tables in documents using OpenCV and object detection algorithms like vision transformer.

I worked on another aspect of LLM training, namely Reinforcement Learning from Human Feedback, at the renowned US-based Artificial General Intelligence company Turing, where I worked on training models of Apple by writing prompts for the models and evaluating the responses. This experience introduced me to another critical aspect of the LLM training pipeline, where I actively contributed to refining model behavior through human feedback, ensuring alignment with real-world user needs.

I am doing a masters in theoretical physics. My research interest is applications of machine learning in particle physics.