Key Contributor to IDEFICS Model

Played a significant role in the development, fine-tuning, and popularization of IDEFICS (Image-aware Decoder Enhanced à la Flamingo with Interleaved Cross-attentionS), Hugging Face's open-access visual language model, enabling new multimodal AI applications.

Advancements in Parameter-Efficient Fine-Tuning (PEFT)

Actively contributed to the PEFT library, developing and promoting techniques like LoRA (Low-Rank Adaptation) that allow for efficient adaptation of massive pre-trained models with significantly fewer trainable parameters.

Prolific Open-Source Contributor

Consistent and impactful contributor to core Hugging Face libraries such as Transformers, Diffusers, and Accelerate, providing code, bug fixes, examples, and documentation that empower thousands of developers and researchers globally.

Thought Leadership in LLM Fine-Tuning

Authored influential blog posts and guides on fine-tuning Large Language Models, sharing practical insights and best practices that have become valuable resources for the AI community.

Development of Zephyr Models

Contributed to the development and training of Zephyr models, a series of high-performing language models fine-tuned with direct preference optimization (DPO), showcasing advancements in instruction-following and conversational AI.

MBA, Marketing & Finance

SRM University - Year 2008

Fathima Matriculation School