Soroush Saryazdi
Co-Founder at Swerve
Mountain View, California, United States
Summary
🤩
Rockstar🎓
Top SchoolSoroush Saryazdi is an AI and robotics engineer with eight years of experience building ML-driven perception and autonomy systems, now co-founding a Y Combinator-backed startup in Mountain View. He led neural network teams at Matic, designing voxel occupancy and BEV planning models for autonomous consumer robots, and has hands-on research engineering experience in deep learning. An active open-source contributor to gradslam, he implemented core ICP functionality for differentiable dense SLAM in PyTorch, reflecting deep expertise in 3D pointcloud processing and optimization. Soroush combines academic training from Concordia (MS) and engineering roots in electrical and electronics, bringing a researcher’s rigor to production robotic systems. He’s comfortable moving between low-level geometric algorithms and deployable neural architectures, and often bridges research prototypes into scalable product features.
8 years of coding experience
2 years of employment as a software developer
Y CombinatorMaster of Science - MS, Computer Science, Master of Science - MS, Computer Science at Concordia UniversityBachelor of Science - BS, Electrical and Electronics Engineering, Bachelor of Science - BS, Electrical and Electronics Engineering at Shahid Bahonar University of KermanHigh School Diploma, Physics and Mathematics, High School Diploma, Physics and Mathematics at National Organization for Development of Exceptional Talents (Sampad)
Persian, English
Github Skills (10)
pytorch10
machine-learning10
slam10
python10
3d-reconstruction10
algorithms9
data-structure9
algorithm9
data-structures9
computer-vision9
Programming languages (7)
Github contributions (5)
gradslam/gradslam
Mar 2020 - Nov 2020
gradslam is an open source differentiable dense SLAM library for PyTorch
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:1 release, 3 reviews, 49 commits in 8 months
Contributions summary:Soroush restructured the layout of the code, specifically focusing on the `ICP.py` file within the `slamaware2d/odometry` directory. The code changes involve implementing core functionality for Iterative Closest Point (ICP) algorithms. This includes functions for pointcloud transformations, normal transformations, solving linear systems, and Gauss-Newton optimization steps, demonstrating a focus on 3D point cloud processing and SLAM-related computations.
pytorchroboticsdifferentiabledeep-learning3d-reconstruction
saryazdi/Duckietown-Object-Detection-LFV
Dec 2019 - Dec 2019
Contributions:242 commits, 241 pushes, 1 branch in 4 days
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.