About Me
I am a Senior Researcher at Microsoft Research, working on "Physics of Large Language Models". I am broadly interested in enhancing LLMs through new data sources, training regimens, and model architectures.
I have a Ph.D. in Computer Engineering from UC San Diego, advised by Professor Farinaz Koushanfar. My thesis focused on algorithms for the automated design of efficient and robust Deep Learning models in the vision and NLP domains. Please see the "Research" tab for an overview of my research interests. My research has been awarded the 2019 Qualcomm Innovation Fellowship.
News
Phi-3 is announced! The smallest phi-3-mini, 3.8B model is comparable to GPT-3.5 and beats Llama-3 8B on many benchmarks. Technical Report, HF Model
Gave a talk on our Phi series of language models at the 2023 NeurIPS Large Language Model Efficiency Challenge. (link to the video on NeurIPS website)
Phi-2 blog post is out: Phi-2: The surprising power of small language models
Phi-2, our most recent small language model, was announced by Satya Nadella at Microsoft Ignite 2023.
Research
My research delivers solutions that target the efficiency, robustness, accuracy, and privacy of deep learning algorithms in data-intensive, embedded scenarios. A high-level list of my research interests are:
Deep Learning on Constrained Platforms
AutoML and Neural Architecture Search
Hardware/Algorithm Co-design
Robust Deep Learning
Privacy-preserving Deep Learning
Experience
Senior Researcher
MSR Redmond, Jan 2023-present
Part-time Researcher
MSR Redmond, Jan 2022-Jun 2022
Research Intern
MSR Redmond, Jun 2021-Sep 2021
Apple AI Research, Jun 2020 - Sep 2020
Apple Siri Search, Jun 2019 - Sep 2019
Education
Ph.D. in Computer Engineering (2023)
Advisor: Prof. Farinaz Koushanfar
UC San Diego, USA
M.Sc. in Computer Engineering (2019)
Advisor: Prof. Farinaz Koushanfar
UC San Diego, USA
B.Sc. in Electrical Engineering (2017)
Sharif University of Technology, Iran