Vipul Ved Prakash is a software engineer and Internet entrepreneur known for his contributions to open-source software and artificial intelligence. He co-founded the anti-spam company Cloudmark, the social-media search company Topsy, and the open-source AI company Together AI. In the early 2000s, he gained recognition for creating Vipul's Razor, a widely used open-source anti-spam system, and was named to MIT's Technology Review list of the Top 100 young innovators in the world in 2003 at age 25.
Prakash's career spans multiple technology domains, from Internet privacy and anti-spam systems to social media analytics and Generative AI. His work has consistently emphasized open-source principles and practical applications of emerging technologies.
Early Life and Education
Prakash grew up in New Delhi, India, where he developed interests in both academics and athletics, playing competitive table tennis. He attended St. Stephen's College, Delhi for undergraduate studies in mathematics, physics, and computer science, but dropped out to pursue his interests in software design. This decision marked the beginning of a career focused on hands-on technical innovation rather than formal academic credentials.
Early Career and Open Source Contributions
In the late 1990s, Prakash co-founded Sense/Net with Ashish Gulhati, an Internet privacy company serving early Internet users in India. During this period, he also wrote for computer magazines and industry journals, including a regular column called "Net Zeppelin" on networking protocols for the Indian edition of PC World Magazine.
Prakash authored several extensions to the Perl programming language, published under open-source licenses and distributed via CPAN. In May 2000, he and Rishab A. Ghosh published the Orbiten Free Software Survey, considered one of the first successful attempts at building an empirical model of contribution to open-source projects. This work demonstrated his early commitment to understanding and improving the open-source ecosystem.
As a self-described cypherpunk, Prakash created a dolphin-shaped implementation of the RSA algorithm in Perl, which was printed on a T-shirt and sold by ThinkGeek as a protest against restrictive crypto export laws. This project combined technical skill with advocacy for digital privacy and freedom.
Cloudmark and Anti-Spam Innovation
Prakash worked as an engineer at Napster, where he met co-founder Jordan Ritter. Together, they co-founded Cloudmark to provide a commercial version of Vipul's Razor, Prakash's open-source anti-spam system. The technology used collaborative filtering and distributed computing to identify and block spam emails, becoming a significant tool in the early 2000s fight against email abuse. His work on this system earned him recognition from MIT's Technology Review in 2003.
Topsy and Apple
Prakash later co-founded Topsy, a social-media search company that indexed and analyzed conversations on platforms like Twitter. In 2013, Topsy was acquired by Apple Inc., with the acquisition reported to be valued at over $200 million. Following the acquisition, Prakash served as a Director of Engineering at Apple, where he led a growing search program. He appeared on the WWDC 2015 stage to introduce iOS Search APIs, highlighting his role in bringing advanced search capabilities to Apple's mobile operating system.
Together AI and Current Work
As of 2026, Prakash is CEO of Together AI, an open-source AI company focused on developing and deploying Large language models and related infrastructure. The company aims to provide accessible tools for Machine learning research and applications, building on Prakash's long history of open-source innovation. Together AI operates in the rapidly evolving field of Artificial intelligence, competing with other major players in the space.
Throughout his career, Prakash has maintained a focus on practical, user-oriented technology solutions, from anti-spam systems to AI infrastructure. His trajectory reflects the evolution of the Internet industry from its early days to the current era of Deep learning and Transformer (architecture)-based models.