NATHAN GOLLAY
Compassionate ∩ Curious ∩ Persistent ∩ Innovative ∩ Collaborative
boX: Custom Deep Learning Server
An Ubuntu Server Configuration for Remote Deep Learning Tasks
Summary
About 8 years old now, this is by far the oldest project on my portfolio. The hardware of this project is close to outdated, but still very powerful for smaller deep learning models. I built this server from hand-picked components, configured the BIOS, and installed Ubuntu Server. Anyone who has worked with NVIDA on a Linux based OS knows what comes next… configuring the driver for the graphics card, which takes a calm mind and a few prayers.
Piecing together conflicting documentation and hurdling over intentional stumbling blocks in NVIDIA’s configuration, I was able to set up this server with parallel computation with compatible versions of Tensorflow and PyTorch- as well as Numpy and so on. The system has served as my primary ML research environment, supporting experimentation across computer vision, physics-informed ML, and audio systems.
I also configured the server for remote connections from all over the world so that no matter where I was, I could run deep learning models- even in India.
Key Contribution: Designed, built, and configured a custom GPU compute server for deep learning research, establishing a stable Ubuntu-based environment with NVIDIA GPU drivers, CUDA-compatible TensorFlow and PyTorch stacks, and remote access for running experiments from anywhere.
Skills
End-to-End Ownership of Research Tooling ▪︎ GPU-Accelerated Machine Learning Environments ▪︎ TensorFlow & PyTorch Environment Management ▪︎ Rapid ML Experimentation Infrastructure
You may also be interested in:
‘Deep Gerchberg-Saxton’: A Physics Informed Neural Network Architecture
A novel deep learning approach to Volumetric Phase Retrieval problems; applied to immersive displays
F.R.I.D.A
A Star Wars-inspired animatronic with face tracking eyes and LLM-enabled communication
Ocean Site One
Lead Engineer for an International Event, Turning a Large Format 360° Video into a Touchscreen-Like Explorable World Using Computer Vision
Fluid Simulation OpenGL Engine
OpenGL volume rendering engine for investigating acoustic fields in density dependent immersive displays.
