amina@aminarhe.me:~$ whoami
I’m a software engineer building across AI/ML and robotics, from GPU clusters and LLM serving to robot learning and hardware. I learn new fields fast, but spend most of my time understanding the problem from first principles, down to the math. I’m drawn to ambitious problems, like automating synthetic biology or building technology for AI treaties.
![a pcb, designed by me [rhyme!]](/_astro/02-trilo-pcb.Bo63DgVV_Z1sGlmG.webp)
amina@aminarhe.me:~$ cat drives.txt
I believe technology should be accessible, inclusive, and serve the goal of making the world a better place.
The problem I can’t stop thinking about is AI risk, and I want to work on it, in AI safety or adjacent to it.
I’m also into philosophy, history, and art for social change. Hot takes welcome by email.
amina@aminarhe.me:~$ cat experience.md
Infrastructure and Software Engineer Intern, May to August 2026
Built a heterogeneous 16-GPU cluster (H100 and A100) under one Slurm-on-Kubernetes scheduler, plus its CI/CD pipeline. Deployment is reproducible through GitOps: a separate dev cluster came up from the same declarative config with no manual setup.
stack: Kubernetes, Slurm, FastAPI, PostgreSQL, vLLM, GitHub Actions, React
Three internships, 2024 to 2026
Model predictive control for robotic-arm motion planning: no overshoot, about 2× faster settling, and 96–99.7% lower steady-state error.
stack: Model Predictive Control, MuJoCo, Python, C++, Optimization
Building software for robotic lab-automation platform, to run biology protocols on the robot: multichannel pipetting, cross-contamination modeling, and labware layout.
stack: TypeScript, Python, Google OR-Tools, constraint programming
Prototyped the first camera system for Trilobio robot arm, with no vision sensing before, end to end across mechanical, electrical, and software engineering.
stack: OpenCV, PCB design, firmware, CAD, CAN, USB 2.0, Oscilloscope
AI research engineering at Jubo Health: a real-time healthcare analytics endpoint over medical records, demoed at the Healthcare+ EXPO. Software engineering at Brothers On The Rise: a web app that automates user-feedback processing.
amina@aminarhe.me:~$ ls Projects/
Worlds in Action Hack
Memories and data become a walkable 3D mind palace: Gemini picks high-signal memories, World Labs renders a navigable Gaussian-splat room in WebXR, and every object maps back to a real note with PICO.
stack: Python, Gemini, World Labs, WebXR
RoboHacks
A multimodal embodied agent on the MARS Innate robot: an ACT imitation-learning policy for manipulation, including handling cutlery, with Gemini for task planning and visual verification and ElevenLabs for voice.
stack: Python, ACT, Gemini, ElevenLabs, Robotics
links: video
NVIDIA Spark Hack
A GPU-accelerated, city-scale earthquake evacuation simulation with NVIDIA RAPIDS: seismic risk, debris propagation, and real-time pathfinding for many agents across USGS and San Francisco geospatial data.
stack: Python, RAPIDS (cuDF, cuSpatial, CuPy), GPU Computing
links: github
NASA Space Apps Challenge, winner in MV, team of 6
A wildfire emergency response system that processes real-time satellite data for adaptive evacuation routing, with IoT hardware in streetlights for emergency communication.
links: project page
Founder, EdTech startup, team of 7
A gamified platform for learning to code, because public-school IT classes in Kazakhstan had little to do with real programming. Raised $6.4K and made 25+ course sales with JavaScript and Python curricula.
stack: Python, JavaScript, Entrepreneurship
World Robot Olympiad, 3rd place, team of 3
In summer 2020 our local Ishim river turned green from an algae bloom. We extracted algae oil for biofuel, then built a photobioreactor and robot to automate it: nutrient dosing, aeration, and light cycles driven by growth-curve data.
stack: Research, Robotics, Science
links: project file
amina@aminarhe.me:~$ cat Research/*.md
with Prof. Carlos Galeano-Ríos (advisor), Aug 2026 to May 2027
Building a Python solver for the dynamic contact of an inflated membrane with a rigid object, using the Kinematic Match free-boundary method: replicating published results, then extending the method to pressurized spherical membranes.
with Prof. Carlos Galeano-Ríos, Cynthia Bortolotto, PhD, and Dmytro Antonovych
Contributed the data visualization and early code implementation to a project developing the Self-Perceived Dissimilarity (SPD) metric, which quantifies how dissimilar a dataset is from each element’s perspective. The team proved its mathematical properties, including semi-metric behavior and global maximization conditions for well-separated clusters. Applications include balanced team formation, clinical trial design, and clustering validation.
amina@aminarhe.me:~$ ls ~/Elsewhere
I lived in six countries, four continents, mostly through Minerva. Each one stretched my thinking, challenged my biases, and sharpened my non-verbal communication and survival skills. Each also taught me a sport.
# type help, or just keep scrolling
