๐ | As of September 2021, I transitioned all my codebases to this GitHub account.
Repositories previously hosted on other version control systems (e.g., GitHub Enterprise, GitLab, Bitbucket) are being gradually migrated here.
๐ | More codebases, documentation, and research toolkits will continue to be added over time.
๐ | This account also hosts shared packages from my personal website and the source codes that I developed and shared on MathWorks File Exchange 
I'm Shahrokh (/สษหยทhroสk/, pronounced โSHAAH-rokhโ, Persian: ุดุงูุฑุฎ), a Research Scientist and Senior AI/ML Engineer at Google, specializing in deep learning, retrieval and ranking, and generative AI.
- My work focuses on embedding models, multimodal architectures, and large-scale model serving.
- I hold a Ph.D. in Computer Science / Computational Science and Engineering (CSE) from the Georgia Institute of Technology, where I also served as a postdoctoral researcher in the College of Computing
- Before Georgia Tech, I earned an M.Sc. in Computer Science from Imperial College London and another M.Sc. in Structural Engineering from the University of Tehran, where I also completed my B.Sc. in Engineering.
My academic and professional interests span across:
- Deep Learning, Generative AI, and Agentic Systems
- Retrieval, Ranking, and Recommender Models
- Embedding Alignment and Model Calibration
- Scalable Inference, GPU/TPU Model Serving
- Computational Modeling and Reservoir Computing
- Coding has been part of my life since my early school years โ starting with QBasic and Assembly,
and later expanding into Python, C++, Java, MATLAB, JS, Julia, and others. - Iโve developed numerous open-source projects and research toolkits across simulation, AI, and data-driven modeling.
- You can find more about my academic background and professional work on my personal website.
Iโm currently focusing on:
- Advancing embedding-based retrieval and ranking systems at scale.
- Researching agentic AI frameworks and their real-world applications.
- Open-sourcing selected personally developed tools and research codebases.
