Hi, my name is Sina. Currently: Robotics MSE @ UPenn Engineering, ex-SWE at BNY, graduated w/ bachelors in Math + CS from Rutgers University in December 2024

I am really interested in how stuff works across fields like scientific computing, robotics, AI, crypto, web-tech, etc. A lot of my repos are me implementing stuff to see the internals of how it works.

Hopefully, later on in my career I'll proudly publish some open source projects that contribute to improving the quality life for humanity (or some subset of it), but for now I am still learning and exploring by my whim!

Please reach out to me if you're interested in collaborating on something or if you'd like me to work on any Open-Source projects that you're involved in. Email me: shazeghian@gmail.com | LinkedIn

TLDR: to learn more about me go to my website here: https://sina-haz.github.io/

1. Tensegrity Work:

• new repo: https://github.com/Sina-Haz/Tensegrity-GNN-Simulator
• full new tensegrity GNN simulator implementation + short README and how it works done in python/JAX/Jraph, implementing for this paper (https://arxiv.org/abs/2410.12216) + later used fully learned simulator to model partially observable state in CableRobotGraphSim (https://arxiv.org/abs/2602.21331) which went for oral presentation @ L4DC 2026 (was no longer part of group when this came out)
• old repo: https://github.com/Sina-Haz/Tensegrity-Project
• features some failed / unpromising attempts using Nvidia Warp, Taichi, C++ (usually just wasn't worth the effort to implement hybrid analytical + learned simulator in any of these frameworks)

1. Advanced Robotics @ Rutgers (2024)

• Kinodynamic-RRT (https://github.com/Sina-Haz/Kinodynamic-RRT): kinodynamic RRT planner in MuJoCo for systems with dynamic constraints.
• Policy-Gradients-and-MuJoCo (https://github.com/Sina-Haz/Policy-Gradients-and-MuJoCo): Actor-Critic policy gradient algorithm trained on a custom MDP in MuJoCo.

2. Computational Robotics @ Rutgers (2023)

• Robotics-1 (https://github.com/Sina-Haz/Robotics-1): 2D motion planning/collision-checking from scratch — SAT-based collision detection, Minkowski sums, and C-space construction for a rigid body and 2-link arm.
• Robotics-2 (https://github.com/Sina-Haz/Robotics-2): sampling-based motion planning (RRT, PRM, PRM*) for a 2-link arm, rigid body, and kinematic car, plus control-blossom RRT under dynamics.
• Robotics-3 (https://github.com/Sina-Haz/Robotics-3): probabilistic localization for a differential-drive robot — dead reckoning, particle filter (MCL), and EKF.

3. Artificial Intelligence @ Rutgers (2023)

• AI-Project-1 (https://github.com/Sina-Haz/AI-Project-1): Python notebook simulating a ship-navigation game with bots, fire-spreading, and path-finding algorithms.

4. Systems Programming @ Rutgers (2023)

• Tic-Tac-Toe-Server (https://github.com/Sina-Haz/Tic-Tac-Toe-Server): multithreaded Tic-Tac-Toe server handling concurrent games, board state tracking, and move/draw command processing.
• Systems-Programming (https://github.com/Sina-Haz/Systems-Programming): collection of systems programming projects covering concurrency, file I/O, and basic OS utilities.

5. Computer Architecture @ Rutgers (2022)

• Computer-Architecture-Projects (https://github.com/Sina-Haz/Computer-Architecture-Projects): collection of computer architecture projects with implementations, test cases, and autograder.
• DAG-SSSP (https://github.com/Sina-Haz/DAG-SSSP): C implementation of single-source shortest path using directed-acyclic-graph data structure.
• NYC-Pathfinder (https://github.com/Sina-Haz/NYC-Pathfinder): Dijkstra-based shortest-path finder for weighted NYC street graph.
• Cache-Simulator (https://github.com/Sina-Haz/Cache-Simulator): C cache simulator modeling 2-level exclusive cache with configurable associativity and FIFO/LRU policies.

1. SLAM (https://github.com/Sina-Haz/SLAM-Experiments)

• studying SLAM from this book: slambook-en (https://github.com/gaoxiang12/slambook-en) w/ both code + math and articles on my website

2. fast.ai course (https://github.com/Sina-Haz/fastai-kaggle) & Part 2: Stable Diffusion (https://github.com/Sina-Haz/fastai-pt-2)

• working through fast.ai's deep learning course, part 1 (practical DL) through part 2 (foundations to stable diffusion)

3. Generative Modelling (https://github.com/Sina-Haz/Generative-Modelling): foray into generative modelling with clean, readable implementations to solidify understanding of the topics.
4. RL (https://github.com/Sina-Haz/RL): hands-on coding projects for the Hugging Face RL course, including model implementations and utilities.
5. NameMe (https://github.com/Sina-Haz/NameMe) & nanoGPT-Plus (https://github.com/Sina-Haz/nanoGPT-Plus): first experiment into language-model development, and extending Karpathy's nanoGPT with custom enhancements.
6. HTTP-Server (https://github.com/Sina-Haz/HTTP-Server): from-scratch HTTP server written in Go, teaching low-level networking and server logic.
7. Vectorgrad (https://github.com/Sina-Haz/Vectorgrad): mini PyTorch-style autograd framework implemented in Python, providing vector-calculus gradients for CPU-based deep-learning tasks.