RACHIT AGRAWAL

Developed and fine-tuned transformer-based LLMs (BERT, ROBERTa) for commonsense question answering using PyTorch and Hugging Face, advancing model performance for complex NLP challenges.

Implemented knowledge distillation pipelines to transfer performance from large teacher models to smaller student models, optimizing accuracy while significantly reducing inference cost.

Built a local Flask web interface with Bootstrap integration to serve LLMs, enabling real-time predictions and history tracking across various NLP tasks.

Gained hands-on experience in programming neural networks using TensorFlow for real-world image processing tasks, enhancing practical AI/ML application skills.

Earned multiple Google Developer badges, including Product Image Search, Object Detection, and Image Classification, demonstrating validated applied AI/ML knowledge and expertise.