Howdy, Samir here
Software Engineer
I like to build autonomous systems
From my educational background, I designed, manufactured, and analyzed mechanical systems. During my journey, I stumbled across the field of robotics. At first, I thought it was cool to build the physical hardware (I still do), but what really peaked my interest was the software behind the autonomous behaviors.
This interest has led me to develop autonomy software for robots that have traversed through land, air, and sea. Most of my contributions and expertise is in GNC (Guidance, Navigation, and Control). Lately, my focus has been more upstream where I get to be more involved with perception and mapping, and I hope to stay in this realm.
Oct 2021 - Feb 2024
I expanded Greensea's autonomy stack for a fleet of underwater robotics. This entailed localization, mapping, and obstacle avoidance for both holonomic and nonholonomic vehicles. Some projects I contributed towards are the following: EverClean (ship hull cleaning service), OPENSEA (hardware agnostic software for underwater robotics), Bayonet AUGV (sea floor surveying), and EOD (Explosive Ordnance Neutralization).
May 2020 - Sep 2021
The main problem we were solving was related to multi-agent systems where we needed decisive coordination between multiple ground and air vehicles. I concentrated on sensor processing, mapping, and navigation for off-road ground automobiles.
Aug 2018 - May 2020
I designed and implemented a data-driven nonlinear longitudinal and lateral controller for ground vehicles.
Aug 2017 - May 2018
I designed, developed, and applied controls for a commercial vehicle. The team ultimately won 2nd place overall.
A Generative Pretrained Model (GPT) using a similar decoder architecture from "Attention is All You Need" utilizes scripts from "The Office" to create scripts similar to the show. There are two tokenization method: character-wise tokens and GPT-2 tokens. While this was tested and trained on a singular GPU, I hope to move this project forward with multi-GPU processing.
Currently, this includes some basic downstream algorithms (system modeling, path tracking, and controls). The goal is to create algorithms for majority of the autonomy pipeline from sensor data processing to controls.
I'm not a frontend developer by any means, but I created this portfolio website using Brittany Chiang's website as inspiration. I learned and executed ideas in an unfamiliar domain.
Brittany Chiang's website design was my inspiration. Next.js, Tailwind CSS, and shadcn/ui were used to build this portfolio website.