At Immverse Al, I contributed to the development and fine-tuning of BhartiyaGPT, a specialized large language model. I integrated LLM APIs and optimized prompts to improve response accuracy and consistency across use cases.
I also developed an automated Q&A generation pipeline to extract structured question-answer pairs from raw data. My work included web scraping, data cleaning, and preparing datasets for fine-tuning the company's internal LLM.
In my LungCancerPrediction project, I developed a CNN and transfer-learning system to classify four cancer types from CT scan images, achieving approximately 87% accuracy. I implemented the data pipeline and model training with TensorFlow/Keras.
I built PDFAI, a RAG web app for querying uploaded PDFs, and an automated EDA pipeline that uses a local LLM to generate analytical insights. I work with Python, SQL, machine learning, and LLM tools.
