Leeroo
Trainable Deep Agents
Leeroo develops trainable deep agents. They are "deep" because they perform complex reasoning and dynamic information integration during inference, and "trainable" because they continuously adapt through user feedback—capturing subtle nuances and personalized requirements unique to each task and domain.
Active Founders
Alireza Mohammadshahi, Founder
Co-founder and Chief Scientist at Leeroo, hold a PhD from EPFL. Previously, I developed an LLM-based question generation and evaluation systems at Meta AI and a compact massively multilingual translation model at NAVERLABS, both with more than 0.5 million downloads, widely benefiting the AI community.
Arshad Shaikh, Founder
I am the Co-founder and CTO of Leeroo, bridging the gap between research and production. Previously, I shaped scalable knowledge-tracing recommender systems in edtech, impacting millions of users. I also contributed to AI research at IIT Bombay.
Majid Yazdani, Founder
I am the co-founder and CEO of Leeroo. Before that, I was a scientist at Meta AI, Linkedin AI, the University of Geneva, and Idiap Research Institute. I got my PhD in machine learning from EPFL.
