Dheeraj Pai
CEO at Leanmcp | AI/ML Researcher | Entrepreneur in Residence
pai
San Francisco, California
Pittsburgh, Pennsylvania
Joined June 2025
Network
8.5K connectionsYCMACW
ILCCAB
MKRDLN
BRDSMA
YSJHTC
BBAJ
TESD
Summary
Dheeraj Pai is an emerging leader in AI and ML, particularly in the domain of large language models and multi-modal AI. As the Founder and CEO of Leanmcp, he is driving the development of the Model Context Protocol (MCP), a platform for building and deploying AI agents and working on open-source AI standards. His work is aimed at improving the interaction between machines and humans by enabling AI to comprehend visual, textual, and auditory data. github+2
His academic background is strong, with a Master of Science in Artificial Intelligence Engineering - Information Security (MSAIE-IS) from Carnegie Mellon University and a dual Bachelor of Technology in Electrical and Computer Engineering from the Indian Institute of Technology Madras. This combination of degrees highlights his expertise in both core computer science and the specialized field of AI, particularly at the intersection with security concerns. github+1
Dheeraj has a track record of entrepreneurship, co-founding Hyperweb AI, an AI and ML startup focused on the educational sector, and currently leading Leanmcp. He also serves as an Entrepreneur in Residence at Antler, indicating his involvement in the startup ecosystem. Furthermore, he is noted as a winner of multiple Olympiads in cryptography and cybersecurity. ynos+1
He is also an active researcher and contributor to the AI community, with several publications in areas like conversational agents, multimodal QA, and neural decoders for topological codes. He has also served as a Graduate Teaching Assistant for Deep Learning at CMU, demonstrating his commitment to education and mentorship in the field. github+2
Work
Education
Projects
Writing
An Interactive Conversational Agent to Aid Human Learning
January 1, 2023Introduces an intelligent agent designed to assist human learning processes.
Neural decoder for topological codes using pseudo-inverse of parity check matrix
January 1, 2019Details a neural network-based decoding method for topological codes utilizing the pseudo-inverse of the parity check matrix.
Neural Network Based Decoder for Topological Codes
January 1, 2019Multihop Multimodal QA using Joint Attentive Training and Hierarchical Attentive Vision Language transformers
Explores methodologies for multi-hop, multi-modal question answering using joint attentive training and hierarchical attentive vision-language transformers.
Tip-of-the-Tongue Retrieval leveraging Large Language Models
Investigates how large language models can be used for "tip-of-the-tongue" retrieval, a cognitive phenomenon where information is known but temporarily inaccessible.