I have been working in the space of NLP for the last 11 years, beginning from my Master's at LTI at Carnegie Mellon University, and then in the industry at Microsoft And Meta. Spanning the range of classical ML models to new-generation LLMs, my focus areas have been Conversational & multimodal AI, data efficiency of generation models, LLM evaluation and safety. Some of the product areas I have worked on are: virtual assistants, Reality Labs and Llama models. I am driven by understanding the mathematics behind the models and why they behave the way they behave.