Shahrokh Shahi

Shahrokh (/ʃɑ:hroʊk/) is a Research Scientist and Senior AI/ML Engineer at Google, specializing in deep learning and generative AI, with extensive work on embedding models and retrieval & ranking systems powering products within Vertex AI Search.

• At Google, Shahrokh actively collaborates with Google DeepMind on the development of embedding models including the Gemini Embedding models, leading efforts that bridge research innovation and large-scale deployment. He also led the productionization of embedding and cross-attention models within Vertex AI Search, improving the quality of search, ranking, and recommendation systems across multiple product verticals.

• More recently, he has been significantly focused on the design and development of agentic AI systems contributing to launching Agentspace and Gemini Enterprise—Google’s large-scale initiatives advancing agentic intelligence and applied AI workflows.

• Dr. Shahi earned his Ph.D. in Computer Science (CS) / Computational Science and Engineering (CSE) from Georgia Institute of Technology, where he also held a Postdoctoral Researcher position in the College of Computing under the mentorship of Elizabeth M. Cherry.

• He also holds multiple M.Sc. degrees across diverse disciplines in engineering and computing, including Computer Science from Imperial College London, where he conducted research under the supervision of Alessandra Russo and Mahdi Cheraghchi. He earned his B.Sc. and M.Sc. in Engineering from the University of Tehran, where he conducted research at the High Performance Computing Lab advised by Soheil Mohammadi.

• His academic and professional work bridges theoretical research and scalable applied AI, with interests spanning LLMs, multimodal systems, generative modeling, agentic frameworks, and trustworthy AI.

The ongoing projects, publications, and open-source computational packages are available on this website. The teaching portfolio also includes material from courses he has taught, including Georgia Tech’s renowned Design and Analysis of Algorithms, featuring lecture notes, recordings, and assignments.